Reducing Air Pollution from Urban Transport - …documents.worldbank.org/curated/en/989711468328204490/...Reducing Air Pollution from Urban Transport Ken Gwilliam, Masami Kojima, and

Reducing AirPollution fromUrban Transport

Ken Gwilliam, Masami Kojima, andTodd Johnson

THE WORLD BANK

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Administrator

30425

Copyright © 2004

The International Bank for Reconstruction

and Development/THE WORLD BANK

1818 H Street, N.W.

Washington, D.C. 20433, U.S.A.

Telephone 202-473-1000

Internet www.worldbank.org

All rights reserved

Manufactured in the United States of America

Published June 2004

The findings, interpretations, and conclusions expressed here do not necessarily reflect the views ofthe Board of Executive Directors of the World Bank or the governments they represent. The WorldBank cannot guarantee the accuracy of the data included in this publication and accepts no responsi-bility for any consequence of their use. The boundaries, colors, denominations, and other informationshown on any map in this work do not imply on the part of the World Bank any judgment of the legalstatus of any territory or the endorsement or acceptance of such boundaries.

Cover: Todd Johnson, 2004; Shanghai, China.

iii

Contents

Acknowledgments .................................................................................................. ix

Foreword ...................................................................................................................xi

List of Abbreviations, Acronyms, and Glossary ............................................ xiii

Preface ...................................................................................................................... xv

Executive Summary ............................................................................................ xviiBackground ................................................................................................... xviiA Framework for Decisionmaking ............................................................ xviiPolicy Instruments for Reducing Transport Emissions and Reducing

Human Exposure .................................................................................. xviiiConclusions ....................................................................................................xxv

1. The Context of the Problem ........................................................................... 1The Air Quality Problem in Developing Countries ...................................... 1Transport as a Source of Pollution .................................................................. 1Air Pollution Levels and Trends ...................................................................... 2Global Climate Change ..................................................................................... 3Urban Transport Policy in Developing Countries ........................................ 4The Policy Stance ............................................................................................... 5

2. A Systematic Approach to Controlling Urban Air Pollution fromMobile Sources ................................................................................................. 7A Framework for Analysis ............................................................................... 7Air Quality Monitoring and Standards .......................................................... 7The Determinants of Transport Emissions ..................................................... 8Assessing Air Pollution Mitigation Measures ............................................. 11Cost-Effectiveness Analysis ........................................................................... 15The Results of Analysis of Air Pollution Control ........................................ 17Appraising Instruments: A Structure for Policy Appraisal ....................... 20

3. Reducing Emissions per Unit of Fuel Consumed .................................... 23Cleaner Fuels .................................................................................................... 23Maintaining Fuel Standards ........................................................................... 32Alternative Fuels ............................................................................................. 34Vehicle Technology .......................................................................................... 40Vehicle Replacement Strategies ..................................................................... 48

4. Reducing Fuel Consumption per Unit of Movement .............................. 53Improving Fuel Efficiency through Vehicle Technology ............................ 53Increasing Fuel Efficiency through Vehicle Operation ............................... 54Encouraging Nonmotorized Transport ........................................................ 55Regulation and Control of Public Road Passenger Transport ................... 59The Role of Mass Transit ................................................................................. 63

5. Reducing Total Transport Demand ............................................................. 65Land Use Policy ............................................................................................... 65Road Pricing ..................................................................................................... 67Physical Restraint Policies .............................................................................. 69

iv REDUCING AIR POLLUTION FROM URBAN TRANSPORT

Parking Policies ................................................................................................ 69The Special Problem of Motorcycles ............................................................. 70

6. Designing a Supportive Fiscal Framework ............................................... 71Direct Taxation on Emissions ......................................................................... 71Fuel Taxation .................................................................................................... 71Taxation on Vehicles ........................................................................................ 74Constructing a Road Transport Tax Package ............................................... 74Property Taxation and Fees ............................................................................ 75Public Expenditure Policies ............................................................................ 75

7. The Supporting Institutional Framework ................................................. 77The Range of Institutions Involved in Urban Air Quality ......................... 77The Role of Central Government .................................................................. 77The Hierarchy of Government and Inter-Jurisdictional Collaboration ... 78The Organization of Municipal Government .............................................. 79Involving the Private Sector ........................................................................... 80Nongovernmental Organizations and Civil Society ................................... 82

8. Synopsis: Constructing an Effective Package of Measures .................... 85Adopting a Positive Policy Stance ................................................................ 85Direct Policy Tools ........................................................................................... 86Indirect Policy Tools ........................................................................................ 88Political and Technical Consistency .............................................................. 89“Horses for Courses” ...................................................................................... 90Conclusion ........................................................................................................ 92

Annex 1. Conventional Fuel Technology ......................................................... 93Gasoline Quality Improvement ..................................................................... 93Diesel Quality Improvement.......................................................................... 95Impact on the Refining Industry ................................................................... 99

Annex 2. Trends in Vehicular Emission Standards and FuelSpecifications in the United States ........................................................... 103Clean Air Act Amendments of 1990 ........................................................... 103Air Quality Improvement Research Program ........................................... 106Tier 1 and Tier 2 Emission Standards ......................................................... 107

Annex 3. Trends in Vehicular Emission Standards and FuelSpecifications in the European Union ..................................................... 113European Auto-Oil Programme .................................................................. 113Current and Future Standards ..................................................................... 114

Annex 4. World-Wide Fuel Charter ................................................................. 119

Annex 5. Two- and Three-Wheelers ................................................................ 121Relationships between Mass Emissions and Vehicle and Fuel/

Lubricant Technology ............................................................................ 122Emission Standards for Two- and Three-Wheel Vehicles ........................ 125Controlling Emissions from Two- and Three-Wheelers ........................... 127

Annex 6. Alternative Fuels ................................................................................ 129Natural Gas .................................................................................................... 129Liquefied Petroleum Gas .............................................................................. 133Electric/Hybrid ............................................................................................. 134Biofuels ............................................................................................................ 135Hydrogen and Fuel Cell Technology .......................................................... 137

CONTENTS v

Annex 7. Maintaining Vehicles: Inspection and Maintenance Programs .. 139Data on Vehicle Population .......................................................................... 139Test Procedures .............................................................................................. 139Administrative Control ................................................................................. 144Experience in Mexico City ............................................................................ 148

Annex 8. Estimating the Health Impacts of Air Pollution .......................... 151Selecting the Health Effects to Be Studied ................................................. 151How Are Health Effects Estimated? ........................................................... 151Results from Existing Studies ...................................................................... 154Estimating Health Effects in Developing Countries ................................. 154Conclusions .................................................................................................... 155

Annex 9. Valuing Health Effects ...................................................................... 157Valuing Reductions in Illness ....................................................................... 157Valuing Reductions in Premature Mortality .............................................. 158Valuing Health Benefits in Developing Countries .................................... 159The Policy Relevance of Health-Benefits Analysis—Example from

Mexico City ............................................................................................. 159The Use of Benefit Estimates in Cost-Benefit Analyses ............................ 160Conclusions .................................................................................................... 161

References ............................................................................................................. 163

Tables1 Contribution of Vehicle Exhaust to Ambient Particulate

Concentrations ..................................................................................... 142 Estimated Cost of Air Quality Mitigation Measures ...................... 183 Bus Priority Measures in London ...................................................... 63

A1.1 Gasoline Fuel Parameters that Affect Air Quality .......................... 94A1.2 Diesel Parameters that Affect Air Quality ....................................... 96A2.1 U.S. Industry Average Baseline Gasoline, 1990 ............................. 104A2.2 Simple Model, 1 January 1995–31 December 1997 ....................... 104A2.3 Complex Model, 1 January 1998–31 December 1999 ................... 105A2.4 Federal Diesel Standards .................................................................. 105A2.5 Tier 1 U.S. Federal Exhaust Emission Standards for Light-Duty

Vehicles, Federal Test Procedure, Cold CO (g/km) ...................... 108A2.6 U.S. Federal Heavy-Duty Exhaust Emission Standards

Compression Ignition and Urban Buses (g/kWh) ........................ 108A2.7 Tier 2 FTP Exhaust-Emission Standards for Light-Duty

Vehicles, Light-Duty Trucks, and Medium-Duty PassengerVehicles, Permanent (g/km) ............................................................ 110

A2.8 Phase-in Percentages for Tier 2 Emission Standards for Light-Duty Vehicles, Light-Duty Trucks, and Medium-DutyPassenger Vehicles ............................................................................. 110

A2.9 Gasoline Sulfur Limits in the United States ................................... 110A2.10 Heavy-Duty Gasoline Exhaust Emission Standards for 2004

and Later Model Year ........................................................................ 111A3.1 Automotive Gasoline Specifications in the EU ............................. 115A3.2 On-Road Diesel Specifications in the EU ....................................... 115A3.3 EU Exhaust Emission Standards for Passenger Cars (g/km) ..... 116A3.4 EU Emission Standards for Light Commercial Vehicles (g/km) .. 116A3.5 EU Emission Standards for Heavy-Duty Diesel Engines (g/kWh). 117A3.6 Emission Standards for Diesel and Gas Engines, European

Transient Cycle Test (g/kWh) .......................................................... 117A4.1 World-Wide Fuel Charter Gasoline Specifications ....................... 120

vi REDUCING AIR POLLUTION FROM URBAN TRANSPORT

A4.2 World-Wide Fuel Charter Diesel Specifications ............................ 120A5.1 Independent Variables in Regression Analysis ............................. 123A5.2 Log Particulate Emission Model Specification .............................. 124A5.3 Log HC Emission Model Specification ........................................... 124A5.4 CO Emission Model Specification ................................................... 124A5.5 U.S. Emission Limits for Motorcycles over 50 cc Capacity .......... 125A5.6 Future U.S. Motorcycle Exhaust Emission Standards .................. 125A5.7 ECE Regulation 40/40.01 for Type Approval Exhaust

Emission Limits for Four-Stroke Engine Motorcycles .................. 126A5.8 EU Motorcycle Emission Limits, 1999–2003 .................................. 126A5.9 Common Position on Motorcycle Emission Standards

Adopted in July 2001 by the European Council (g/km) .............. 126A5.10 Type Approval Emission Standards for Gasoline- and Diesel-

Powered Two- and Three-Wheelers in India (g/km) ................... 127A5.11 Emission Standards for New Production Motorcycle Models

in Taiwan, China ................................................................................ 127A6.1 Emissions Benefits of Replacing Conventional Diesel with CNG . 131A6.2 Comparison of CNG and “Clean-Diesel” Buses in New

York (g/km) ........................................................................................ 131A8.1 Human Health Effects of the Common Air Pollutants ................. 152A9.1 Annual Health Benefits due to Ozone and PM10 Reductions

in Mexico City (million 1999 US$) .................................................. 160

FiguresE.1 Factors Contributing to Transport Emissions ................................. xix

1 Sequence of Questions to Appraise Mitigation Options toTackle Mobile Sources ........................................................................... 7

2 Particulate Emissions as a Function of Vehicle Speed .................... 103 U.S. Particulate Emission Standards for Urban Buses ................... 194 Factors Contributing to Transport Emissions .................................. 21

A1.1 Particulate Emissions from New U.S. Heavy-Duty Diesels .......... 95A6.1 Payback for Conversion from Premium Gasoline to CNG in

Argentina, 1999 Fuel Prices .............................................................. 132A7.1 Correlation Between Visible Smoke and Mass Particulate

Emissions ............................................................................................ 142

Boxes1 Actual Levels versus Limits .................................................................... 242 Cost of Fuel Reformulation: Examples from Latin America and

the Caribbean and from Asia .................................................................. 273 Diesel Certification in California .......................................................... 1064 From Dual-Fuel Buses to Dedicated CNG: Lesson from Seattle,

United States ........................................................................................... 1305 Natural Gas Buses: Experience of Bus Fleet Operators .................... 1326 Estimating a Health Impact of Lowering PM10 Concentrations ...... 153

International Experience1 Source Apportionment: Lessons from the United States .................... 122 Diesel Sulfur Contribution to Emissions ............................................... 263 Market-Based Approach to Tackling Abuses in Fuel Markets:

“Pure for Sure” in India ........................................................................... 354 Vehicle Replacement in Bogotá, Colombia, and Delhi, India ............. 525 Bus Rapid Transit in Bogotá .................................................................... 606 Addressing the Environmental Impacts of Bus Competition in

Santiago, Chile .......................................................................................... 617 Congestion Pricing in Developing Countries ....................................... 68

CONTENTS vii

Frequently Asked Questions1 How do you decide when to lower transport fuel sulfur limits,

and to what level? .................................................................................... 292 If it costs only a cent or two a liter to improve fuel quality, why

can’t we improve fuel quality immediately? ........................................ 313 Since CNG produces markedly lower particulate emissions than

diesel, why not promote switching from diesel to CNG in all citieswith serious particulate air pollution? .................................................. 37

4 Biofuels are renewable and hence should surely form animportant component of sustainable transport, so shouldn’t allgovernments actively promote biofuels? .............................................. 39

5 When would it make sense to install passive catalyzedparticulate filters? ..................................................................................... 44

6 Does privatization of public transport lead to worsening urbanair pollution? ............................................................................................. 62

ix

Acknowledgments

This report was commissioned by the Air QualityThematic Group of the World Bank, consisting of spe-cialists from the environment, transport, and energysectors. The report has been approved by the Environ-ment, Transport, and Energy and Mining SectorBoards of the World Bank.

The Air Quality Thematic Group discussed andagreed on the report in detail. Important contributorsto this review process included Ronald Anderson,Asif Faiz, David Hanrahan, Pierre Graftieaux, MagdaLovei, Paul Procee, Richard Scurfield, Jitu Shah, AkiraTanabe, and Robert T. Watson. Nigel Clark, GeorgeBerry Chair of Engineering, Department of Mechani-cal and Aerospace Engineering, University of West

Virginia, conducted a technical review of the first ver-sion of the report in June–July 2003.

Consultation drafts of this report were discussedin workshops in Bangkok, Thailand, in June 2003; RioJaneiro, Brazil, in December 2003; and Washington,D.C., in January 2004. A Web-based consultation wasconducted in March and April of 2004. Commentswere received from national and international nongov-ernmental organizations, academics, industry, and gov-ernments. We are grateful to the participants of the work-shops and all those who provided written commentsduring the consultation process. The authors thankLinda Harteker and Paula Whitacre for editorial assis-tance and Nita Congress for desktop publishing.

xi

Foreword

Urban air pollution from road transport is a growingconcern in a large number of developing country cit-ies. With rising income, the use of motorized trans-port is expected to continue to increase in the comingyears, potentially worsening air quality. Poor air qual-ity in turn has been shown to have seriously adverseeffects on public health. The World Health Organiza-tion estimated that 650,000 people died prematurelyfrom urban air pollution in developing countries in2000.

The need to tackle air pollution from transport iswidely acknowledged. But the menu of options avail-able is varied and can be daunting. Are there keyquestions that should be answered to guide policy-making? Under what conditions are the different miti-gation measures likely to achieve pollution reduction?Are there key steps to be taken or underlying condi-tions that must be met, without which pollution re-duction is unlikely? Which mitigation measures are“proven,” which are more difficult to implement, andwhich are still in the realm of pilot-testing?

This report is intended to provide guidelines andprinciples for answering these and other related ques-tions. The report does not attempt to provide a de-tailed road map applicable to all circumstances—given the varying nature of air pollution, pollutionsources, and available resources, answers and evenkey policy questions will be different from country tocountry—but rather proposes a framework in whichpolicy selection and implementation should occur,drawing lessons from international experience. Itplaces a special emphasis on how to coordinate poli-cies across three sectors most closely linked to themitigation of air pollution from road transport—envi-ronment, transport, and energy—and how to recon-cile the sometimes conflicting objectives and demandsof these sectors to achieve environmental improve-ment.

We hope that this report will stimulate and con-tribute to a discussion on how best to coordinate poli-cies across different sectors to their mutual benefit inan environmentally sustainable manner.

James Warren Evans Maryvonne Plessis-Fraissard Jamal SaghirEnvironment Sector Board Transport Sector Board Energy and Mining Sector Board

xiii

List of Abbreviations, Acronyms, andGlossary

ACEA Association des Constructeurs Européensd’Automobiles (Association of European Auto-mobile Manufacturers)

AQIRP Air Quality Improvement Research Program(U.S. auto/oil industry study)

ARPEL Associacíon Regional de Empresas de Petróleo yGas Natural en Latinoamérica y el Caribe (Re-gional Association of Oil and Natural Gas Com-panies in Latin America and the Caribbean)

ASTM American Society of Testing and Materials

ATC Area traffic control (systems)

BMTA Bangkok Mass Transit Authority

BRT Bus rapid transit

C5 Hydrocarbons with five carbon atoms

CAA Clean Air Act

CAFE Corporate average fuel economy (U.S. stan-dards, set by the Department of Transportation,on the actual sales-weighted average fueleconomy of domestic and imported passengercars and light-duty trucks)

CARB California Air Resources Board

CBD Central business district

CFC Chlorofluorocarbons (refrigerants that have glo-bal warming impacts as well as having damag-ing effects on the stratospheric ozone layers)

CNG Compressed natural gas

CO Carbon monoxide

CO2 Carbon dioxide

COI Cost of illness

CR Concentration-response

CUEDC Composite Urban Emissions Drive Cycle

DALY Disability-adjusted life-year

ECE United Nations Economic Commission for Eu-rope

ECMT European Conference of Ministers of Transport

EEVs Environmentally enhanced vehicles

EGR Exhaust gas recirculation

ELR European load response

EMA Engine Manufacturers Association

EPEFE European Programme on Emissions, Fuels andEngine Technologies (a European auto and oilindustry study)

ERP Electronic road pricing

ESC European stationary cycle

EU European Union

EUDC (EU) Extra urban driving cycle

EWG Environmental Working Group

FTP (U.S.) Federal Test Procedure

GEF Global Environment Facility

GHG Greenhouse gas (gas that contributes to globalwarming effects)

GTZ Deutsche Gesselschaft für TechnischeZusammenarbeit (German Technical Coopera-tion)

GVWR Gross vehicle weight rating

HC Hydrocarbon

HCHO (Molecular formula for) formaldehyde

HDDE Heavy-duty diesel engine

IANGV International Association for Natural Gas Ve-hicles

ICCT International Council on Clean Transportation

I/M Inspection and maintenance (systems)

IQ Intelligence quotient

ITS Intelligent transport systems (computer basedreal-time control systems of traffic or vehicles)

IU In-vehicle unit

JAMA Japan Automobile Manufacturers Association

JASO Japanese Standards Organization

LDT Light-duty truck

LDV Light-duty vehicle

LNG Liquefied natural gas

LPG Liquefied petroleum gas

MCMA Mexico City Metropolitan Area

MON Motor octane number (ability of gasoline to re-sist auto-ignition, or knocking, under highwaydriving conditions)

MTA Metropolitan Transportation Authority

MTBE Methyl tertiary-butyl ether (an oxygenate)

MY Model year

NEPC National Environment Protection Council (ofAustralia)

NG Natural gas

xiv REDUCING AIR POLLUTION FROM URBAN TRANSPORT

NGO Nongovernmental organization

NGV Natural gas vehicle

NMHC Nonmethane hydrocarbons

NMT Nonmotorized transport

NO Nitric oxide

N2O Nitrous oxide (a powerful greenhouse gas)

NO2 Nitrogen dioxide

NOx Oxides of nitrogen

NPRA (U.S.) National Petrochemicals and Refiners As-sociation

NREL National Renewable Energy Laboratory

NTE (U.S.) Not to exceed

OBD On-board diagnostic

OECD Organisation for Economic Co-operation andDevelopment (association of mainly industrialcountries)

OEM Original equipment manufacturer

OLADE Organización Latinoamericana de Energía(Latin American Energy Organization)

PAH Polyaromatic hydrocarbon (a hydrocarbon withmore than one benzene ring)

PM Particulate matter

PM2.5 Particulate matter of size 2.5 microns or smallerin aerodynamic diameter, also referred to as re-spirable particulate matter or fine particulatematter

PM10 Particulate matter of size 10 microns or smallerin aerodynamic diameter, also referred to asinhalable particulate matter

RAD Restricted activity day

RFG Reformulated gasoline

RON Research octane number (ability of gasoline toresist auto-ignition, or knocking, under citydriving conditions)

SAE Society of Automotive Engineers

SET (U.S.) Supplementary emission test

SFTP (U.S.) Supplemental Federal Test Procedure

SO2 Sulfur dioxide

SOx Oxides of sulfur

SPM Suspended particulate matter

STAP Science and Technology Advisory Panel

SUV Sport utility vehicle

TSP Total suspended particles

T90, T95 Temperature at which 90 percent or 95 percentof a fuel evaporates

UNECE United Nations Economic Commission for Europe

URBAIR Urban Air Quality Management Strategy in Asia

USDA U.S. Department of Agriculture

U.S. EPA U.S. Environmental Protection Agency

U.S. GAO U.S. General Accounting Office

VOC Volatile organic compound

VSL Value of a statistical life

VLSY Value of statistical life-years

WHO World Health Organization

WTP Willingness to pay

Units of MeasureB/d Barrels per day

cc Cubic centimeters (a unit of volume)

cSt Centistokes, a unit of kinematic viscosity (vis-cosity divided by density)

€ Euros

g Grams

g/km Grams per kilometer

g/kWh Grams per kilowatt-hour

g/l Grams per liter

kg Kilograms

kg/m³ Kilograms per cubic meter (a unit of density)

km Kilometers

km/h Kilometers per hour

kPa Kilopascals (a unit of pressure)

kW Kilowatts

kWh Kilowatt-hours (a unit of energy)

m Meters

m³ Cubic meters

ppb Parts per billion

ppm Parts per million

psi Pounds per square inch (a unit of pressure; 1 psiis equal to 8.9 kPa)

rpm Revolutions per minute

R$ Brazilian Real

vol% Percent by volume

wt% Percent by weight

wt ppm Parts per million by weight. 10,000 wt ppm is 1percent by weight, 1,000 wt ppm is 0.1 percent,and so on.

µg/m³ Micrograms per cubic meter

µm Microns

xv

Preface

This report is intended to assist World Bank Groupstaff and client countries in the design of appropriatestrategies for controlling the impacts of urban air pol-lution from mobile sources. The report considers onlythe direct air impacts of surface transport, excludingaviation, marine transport,1 non-road vehicles (suchas bulldozers and mining equipment), noise pollu-tion, habitat fragmentation, and waste disposal ofscrapped vehicles. It is aimed at World Bank Groupstaff as well as national and local governmentpolicymakers working in a number of sectors relatedto air pollution from mobile sources—transport, en-ergy, and environment. Main guideline recommenda-

tions and observations appear in bold italics in the

main text.

The report is divided into eight chapters and issupplemented by nine technical annexes and an Ex-ecutive Summary. The Executive Summary providesgeneral guidelines to practitioners and policymakerson key policy themes, cross-references the relevantsections of the report and annexes where the topics arediscussed, and is recommended for those readers whowant an overview of the main messages in the report.

Chapter 1 describes the nature of the problem, thelevels and trends of ambient air pollution in develop-ing country cities, and the context within which trans-port-related air quality policy needs to be set. It em-phasizes that the behavior of the many personal andcorporate actors in the transport sector are fundamen-tal in determining the effectiveness of policy efforts toreduce urban air pollution. Chapter 2 discusses theimpacts of the principal urban air pollutants, and howto assess the contribution of transport to poor urbanair quality. It concludes by identifying three principal

transport aspects within which air quality improve-ment can be sought: through reducing the emission ofpollutants per unit of fuel consumed, reducing theconsumption of fuel per unit of transport services,and limiting the overall demand for motorized trans-port services.

Chapters 3 through 5 discuss, for each of thesethree aspects, the array of policy areas and instru-ments in which improvements can be sought, andidentify the range of instruments that can be used.The annexes supplement these chapters by providingmore detailed information on the physical and eco-nomic characteristics of technologies—both of somecurrent commercially viable technologies and of sometechnologies that are still in development, and alsohow to ensure their proper maintenance—and on theeconomic valuation of health impacts of air pollution.The basis for the guidance and policy discussion inthese chapters draws heavily on the experience of theWorld Bank in the urban transport, fuel, and environ-ment sectors in developing countries. Emphasis is puton the need for solutions to be both affordable andsustainable.

Many of the technological and policy instrumentsdiscussed cannot be sustained, or will be less effec-tive, unless introduced in the appropriately support-ing fiscal framework, discussed in chapter 6, and in-stitutional setting, discussed in chapter 7. Finally,chapter 8 discusses how to formulate a policy pack-age appropriate for the many different situationsfound in developing countries. It identifies a range ofinstruments that grasp synergies between environ-mental and economic policy dimensions or that havebeen found to be cost-effective over a wide range ofcountry circumstances. However, because technologyis changing very rapidly, both in capability and in pri-vate and public cost, and because affordability varies

1Emissions in harbors, inland waterways, and airports canhave a marked impact on urban air pollution.

xvi REDUCING AIR POLLUTION FROM URBAN TRANSPORT

by country, there is no “magic bullet” to solve allproblems. Hence the report does not prescribe a “one-size-fits-all” list of technological imperatives, butrather concentrates on providing information and astrategy framework with which countries may designand adopt air pollution strategies appropriate to theirown environmental, social, and economic circum-stances.

Given the rapid changes in vehicle and pollutioncontrol technology, one caveat should be noted. Therecommendations and observations concerning ve-

hicle technology, fuel quality, and corresponding stan-dards should be interpreted in the light of the circum-stances prevailing at the time of report publication.Different recommendations and observations relatedto technology and standards will undoubtedly be-come more appropriate in the future. In contrast, rec-ommendations and observations concerning sectorand fiscal policies change much less with time. In-deed, setting appropriate sector and fiscal policies cancreate an enabling environment that facilitates ad-vances in standards and technology.

xvii

Executive Summary

BackgroundAir pollution is a serious problem in many develop-ing country cities. Ambient concentrations of fine par-ticulate matter, which is one of the most damaging airpollutants, are often several times higher in develop-ing country cities compared to those in industrialcountries. The largest human and economic impactsof air pollution are the increased incidence of illnessand premature death that result from human expo-sure to elevated levels of harmful pollutants. Usingdamage to human health as the primary indicator ofthe seriousness of air pollution, the most importanturban air pollutants to control in developing countriesare lead, fine particulate matter, and, in some cities,ozone. Air pollution impacts in developing countriesoften fall disproportionately on the poor, compound-ing the effects of other environmental problems suchas the lack of clean water and sanitation.

While the impacts of urban air pollution havebeen documented in both industrial and developingcountries, for policymaking purposes it is importantto know the relative contribution of mobile sources(cars, trucks, buses, motorcycles), stationary sources(power plants, industry, households), and othersources (construction, re-suspended road dust, biom-ass burning, dust storms). In the transport and trans-port fuel-supply sectors, many actors must be part ofan effective strategy for reducing mobile-source emis-sions. To be effective and sustainable over the longterm, regulatory and policy instruments for reducingtransport emissions must provide incentives for indi-viduals and firms to limit the pollution from existingvehicles and to avoid delay in adopting new andcleaner technologies and fuels. Public and private in-stitutions must be equipped with the resources andthe skills necessary to support measures to controltransport emissions and to evaluate the effectiveness

of such measures. Above all, interventions must becost-effective and affordable in light of the myriad ofother pressing needs in developing country cities.

This report discusses policy, technological, admin-istrative, and economic issues surrounding interven-tions for air quality improvement in developing coun-tries, and provides examples of both successful andunsuccessful actions and approaches to air qualitymanagement. The purpose of the report is to assist na-tional and local government policymakers and profes-sionals identify the roles that the transport, energy,environment, and other related sectors play in urbanair quality management in their particular contextsand to help design cost-effective strategies to controlthe impact of mobile-source emissions. It comple-ments the World Bank’s Pollution Prevention and Abate-

ment Handbook (World Bank 1999), which providesgeneral policy advice on pollution management anddetailed recommendations for addressing pollutionfrom stationary sources.

A Framework for Decisionmaking1

In order to design effective approaches to pollutionmanagement from mobile sources, it is important todiagnose urban air pollution problems, determine theimpact of mobile sources, and identify affordable andsustainable solutions. The first step is to ask how seri-ous outdoor air pollution is in a given city and the na-ture of the pollution problem. This entails monitoringair quality and comparing ambient concentrationswith national air quality standards or, in their ab-sence, internationally recognized health-based airquality guidelines. Pollution reduction measuresshould focus on the most damaging pollutants, basedon the combined impact of high ambient concentra-

1Chapter 2.

xviii REDUCING AIR POLLUTION FROM URBAN TRANSPORT

tions, toxicity, and human exposure. Once the mostdamaging pollutants have been identified, the relativecontribution of mobile sources to the problem shouldbe determined (chapter 2). For some pollutants, suchas lead and carbon monoxide, the transport sector isoften a major contributor, while for fine particulatematter the transport sector is typically one of severalsources of emissions. Where transport activities are amajor contributor to a specific air pollution problem,it is important to determine in what ways these activi-ties can be reduced (chapters 3–5). The instruments bywhich emissions can be reduced need to be identified(chapters 3–6), the effectiveness of different policy in-struments assessed, and institutional arrangementsdeveloped and strengthened (chapter 7) to constructan overall policy package (chapter 8).

Policy Instruments for ReducingTransport Emissions and ReducingHuman ExposureThe contribution of transport to air pollution can beviewed broadly as the product of three factors.2

Air pollution from mobile sources can be de-creased by reducing emissions per unit of fuel,3 con-suming less fuel per passenger- or freight-kilometertraveled,4 or requiring fewer passenger- or freight-ki-lometers.5

Effective interventions for reducing transport-re-lated emissions range from general improvements insector efficiency to specific regulatory, policy, and in-stitutional development, and technical measures.Transport emission reduction strategies target eitherthe transport system as a whole or individual ve-hicles, and they can affect both at the same time. Forexample, changing fuel prices can have an immediate

2The three factors were selected for ease of discussing differ-ent dimensions of transport emissions and are not intended to im-ply that fuels and vehicles should be treated separately and in iso-lation from each other. An important objective remains reducingemissions per passenger- or freight-kilometer traveled, which re-quires treating fuels and vehicles as a joint system.

3Chapter 3.

4Chapter 4.

5Chapter 5.

impact on the use of individual vehicles and, overtime, affect the overall composition of the vehiclefleet.

Many measures taken to reduce transport-relatedair pollution will be suboptimal or ineffective over thelonger term without policy changes in the transportand fuel sectors. While such policy changes will rarelybe made based on environmental concerns alone, it isimportant to recognize that reforms in the urbantransport sector or the oil and gas industry can havean enormous positive effect on reducing transport-re-lated air pollution. Some reforms, such as import lib-eralization for clean fuels, will be national in scope,whereas others, such as urban transport sector re-form, will be local. In both cases, they are likely tohave significant economic and social benefits asidefrom their environmental benefits, and in this senseshould be seen as “no regret” measures.

Reducing Emissions through Transport SystemImprovementTransport sector emissions can be reduced through avariety of changes to the overall transport system: ef-ficiency improvements in the urban transport system,changes in modal shares through infrastructure in-vestments or land use policy, or through fiscal policiesthat can affect fuel and vehicle technology choice, fuelconsumption, and vehicle use.

Traffic management 6 and land use7

Traffic system management is intended to smooth theflow of traffic and enhance mobility, but can also havethe added benefit of reducing emissions and fuel con-sumption. Traffic signal control systems are the mostcommon traffic management instruments to securetraffic flow objectives. Segregation of traffic, includingbus priority systems (such as dedicated bus lanes),can decrease variability of traffic speed, enhancesafety, and, equally important, increase the efficiencyand attractiveness of public transport, resulting in sig-nificantly lower fuel consumption and emissions per

6Chapter 4, Traffic Management.

7Chapter 5, Land Use Policy.

EXECUTIVE SUMMARY xix

Factors Contributing to Transport EmissionsF I G U R E

E. 1

= ××Transport emissions

Emissions per unit of fuel

Total transportservices demanded

Fuel consumption per unit of transport service

passenger-kilometer. One weakness associated withsimply improving traffic flow is that faster traffic flowoften invites more traffic. Thus if traffic demand is notcontrolled in parallel, congestion may be little re-lieved and total emissions may even increase.

Options for traffic demand management includepromoting appropriate land use planning to reducetrip lengths, placing restraints on vehicle movementsthrough parking policies, and location- and time-spe-cific charges or bans on certain categories of vehicles.For the structure of land use, high-population densityand the concentration of employment and retail in acentrally located central business district is likely toencourage public transport and reduce trip length.Traffic management is essential to realizing the poten-tial benefits of good land use planning.

Influencing modal choice8

Positive actions to promote alternatives to private mo-torization are important and have been notably suc-cessful in recent years in Bogotá, Colombia, and othercities. These include encouraging nonmotorizedtransport by building and protecting pedestrian andbicycle paths, as well as policies to promote publictransport. Bus sector reform warrants special atten-tion. Buses are much cheaper than rail mass transitand will continue to play an important role in publictransport in developing countries. Bus transportpolicy affects urban air pollution both directly throughits effects on bus vehicle emissions and indirectly

through its effects on the use of smaller vehicles.Without an effective bus system, mobility is providedby numerous small vehicles—three-wheelers, mini-buses, or private cars—contributing to congestion and

8Chapter 4, Encouraging Nonmotorized Transport, Regula-tion and Control of Public Road Passenger Transport, and TheRole of Mass Transit.

reducing average traffic speeds. Policy must thereforeaim simultaneously to minimize the direct air pollutionimpacts of buses by making them clean and to maxi-mize their indirect benefits by making them suffi-ciently attractive to draw passengers from small ve-hicles to high-occupancy transport vehicles.

Higher standards (for emissions as well as qualityof service) for buses need to be introduced in the con-text of a general policy framework that makes theprovision of clean bus services financially viable forthe supplier and affordable to the user. Otherwise,higher standards may raise costs and inadvertentlyreduce service, causing adverse social and environ-mental consequences. The usual effect is the marketentry of informal operators using smaller, often veryold and polluting, vehicles. During the last decade,

Mixing motorized and nonmotorized transport, as well as public

transport vehicles with cars and other vehicle types, reduces the av-

erage speed of traffic and makes it difficult to establish an effective

bus system.

xx REDUCING AIR POLLUTION FROM URBAN TRANSPORT

conventional bus systems have failed in many coun-tries in Africa, Central Asia, and elsewhere as a resultof keeping fares low while attempting to mandate so-cially desirable service quality. When this happens,

Fiscal policies9

It is typically more efficient and cost-effective to taxpolluting vehicles and fuels than to subsidize cleaneralternatives. This would mean that, ideally, taxes onmore polluting fuels such as conventional dieselshould be raised rather than subsidies given tocleaner fuels such as compressed natural gas (CNG).One successful application of differential fuel tax islevying a higher tax on leaded gasoline than on un-leaded gasoline. Subsidies given to public transportfares are generally not cost-effective as an environ-mental policy. There is strong evidence that up to half

of the subsidy “leaks” to benefit transport industry in-terests rather than passengers. Moreover, private carowners are only marginally sensitive to public trans-port fare levels so that the subsidy is not particularlyeffective in shifting travelers from private to publictransport.

Fuel taxation is the most common tax on transportactivity. It is popular not only because of its ease ofcollection and income-generating properties, but alsobecause of its role as a proxy for road user and envi-ronmental charges. Fuel taxes are effective both in re-ducing motorized travel and encouraging fuel-effi-cient vehicles. Unfortunately, a fuel tax is a relativelyweak proxy for local pollutant emission charges be-cause it fails to reflect the location of emissions as well9Chapter 6.

bus operators face increasing costs and decreasingprofits on account of growing traffic congestion andcompetition from weakly regulated paratransit ve-hicles.

General Guidelines for Improving the Transport System

Any intervention in the transport sector needs to be

part of a favorable transport planning and manage-

ment policy framework. That framework should con-

sider the following:

� Formulating transit-oriented development strate-

gies to reduce trip lengths and concentrate move-

ments on efficient public transport axial routes.

� Conducting air quality audits of all new major

transport infrastructure projects as a required

part of the environmental impact assessment to

determine if the projects will lead to or worsen

exceedances of air quality standards.

� Giving priority to buses in the use of road infra-

structure, and particularly the creation of segre-

gated busway systems, in order to improve and

sustain environmental standards for buses.

� Improving the efficiency of bus operation through

the design of more efficient route networks, better

cost control, and creation of incentives for im-

provement through commercialization and com-

petition.

� Promoting competitive bidding for transport fran-

chises based on performance-based criteria, in-

cluding emission characteristics of vehicles.

� Establishing adequate and safe pedestrian and

bicycle facilities in order to promote nonmotorized

options for short distance trips.

� Establishing and implementing protocols for traffic

signal system settings that result in reduced ex-

haust emissions.

� Establishing urban traffic management centers

and involving police in traffic management system

design and training.

� Establishing a municipal department or agency

with comprehensive responsibility for integrated

land use and transport planning, including envi-

ronmental protection issues.

EXECUTIVE SUMMARY xxi

as the amount of emissions per unit of fuel consumed.For that reason, more precisely targeted alternativesto fuel taxes should be considered in parallel.

Taxes for petroleum fuels should take careful ac-count of, and minimize, the possibilities for fuel adul-teration and socially undesirable inter-fuel substitu-tion. There is a strong case for setting the gasoline taxabove the general tax rate on commodities on distri-butional grounds as well as to direct efficient alloca-

Reducing Emissions at the Vehicle Level10

Improved fuels and vehicle technology have enor-mous potential for reducing vehicle emissions, andfuel and vehicle standards are often the most widelydiscussed policy options for tackling mobile sourceemissions. In this context it is very important to treatfuels and vehicles as a joint system, since cleaner ve-hicle technology generally requires improved fuelquality. The ultimate objective is to adopt a fuel andvehicle system embodying high standards and bestpractice technology that have been proven cost-effec-

tive in the industrial countries. The question is notwhether to adopt these standards in developing coun-tries, but how and when to adopt them. The pace ofthat transition will be determined by the cost-effec-tiveness of such measures to improve air quality com-pared with other measures (including those in othersectors), given the constraints in human and financialresources.

Improving vehicle technology is not sufficient toensure that emissions will remain low over the life-time of the vehicle. The state of vehicle repair isknown to have a great impact on the amount of pollu-tion generated and of fuel consumed. Fuel and ve-hicle technology measures will be most effective in re-ducing emissions if vehicles are routinely repaired

10Chapter 3; chapter 4, Improving Fuel Efficiency through Ve-hicle Technology, Increasing Fuel Efficiency through Vehicle Op-eration; and annexes 1–7 .

tion of resources in developing countries. There is alsoa strong case for a diesel tax as the principal means ofcharging heavy vehicles for wear and tear on the roadand capturing the marginal social damage from dieselemissions. Because of the significant impact of highertaxation on non-automotive uses of diesel—in railtransport, agriculture, and industry, for example—itmay be sensible to give rebates on the higher dieseltax to non-road users.

General Guidelines for Fiscal Policies

The economic ideal would be a system of direct

taxation on emissions, combined with trading of

emission certificates among fuel and vehicle manu-

facturers, but the complexity of such a system

makes it necessary to devise alternatives. Among

the fiscal policies that can be used to reduce trans-

port sector emissions are the following:

� In those countries where taxes on diesel fuel for

transport use are very low, raising them to com-

pensate for environmental damages, pay for road

wear and tear, and encourage fuel-efficient ve-

hicles and the use of cleaner fuels.

� In addition to fuel taxes, considering separate ve-

hicle charges based on vehicle weight, axle load-

ings, and annual mileage.

� Introducing direct charges for the use of urban

road space, including congestion charges.

� Introducing or raising taxes, import duties, and

vehicle licensing disincentives for polluting ve-

hicles and engines.

� Giving serious consideration to eliminating subsi-

dies to public off-street parking as well as not

permitting free on-street parking, especially

where they increase congestion by generating

private transport trips to congested locations, or

where on-street parking increases congestion by

reducing available road space.

xxii REDUCING AIR POLLUTION FROM URBAN TRANSPORT

and serviced, if cheaper but lower-quality counterfeitspare parts are avoided, and if exhaust control de-vices and other parts of the vehicle affecting emis-sions are properly maintained. Fostering a system ofregular and proper preventive vehicle maintenanceamong both public and private vehicle owners is anessential element of urban air quality management.Driving behavior, particularly the avoidance of exces-sive acceleration, is also important to fuel consump-tion and emissions.

Inspection and maintenance11

Vehicle inspection and maintenance (I/M) programscan help improve vehicle maintenance behavior and

enforce emission standards for in-use vehicles. Theprimary objective of I/M systems is to identify grosspolluters and ensure that they are repaired or retired.Test protocols should be designed to minimize falsepasses or false failures, make it difficult to cheat oravoid inspection, minimize measurement differencesamong test centers, and maximize reproducibility andaccuracy. The I/M system in Mexico City is an ex-ample of a successful program on a large scale. Expe-rience in Mexico has shown that high volume, central-ized test-only centers are more effective thandecentralized test-and-repair garages. Given limitedresources available, it may be advisable to concentrateresources on categories of vehicles likely to contain alarge fraction of high annual-kilometer, gross pollut-ers (for example, commercial diesel vehicles), ratherthan test every vehicle each year.11Chapter 3, Vehicle Technology; and annex 7.

Fuel quality12

Conventional liquid transportation fuels will remainthe primary focus of fuel quality improvements forthe foreseeable future. Improving gasoline by refor-mulating it can involve eliminating lead and reducingbenzene, sulfur, vapor pressure, and total aromatics,

General Guidelines for Inspection and Maintenance

An effective vehicle inspection program can help en-

force emission standards for in-use vehicles. Inter-

national experience suggests the following advice

on establishment of an I/M system:

� The government must be willing and able to provide

the resources for auditing and supervising the pro-

gram (even if the supervision is outsourced) that are

needed to guarantee its objectivity and transparency.

� Centralized, test-only private sector centers with

modern instrumentation, maximum automation,

and “blind test” procedures are easier to control

for quality; all centers should be subject to inde-

pendent monitoring.

� An up-to-date and accurate vehicle registration

record is necessary coupled with a requirement

to display a visible sticker certifying that the ve-

hicle has been inspected and passed, under pen-

alty of a fine large enough to deter evasion, to en-

sure that all vehicles in the designated categories

report for testing.

� Education campaigns and clinics to improve ve-

hicle maintenance, especially for two-stroke en-

gine maintenance and lubrication, can be helpful

complements to I/M.

while reformulating diesel properties (including low-ering density, sulfur content, and polycylic hydrocar-bons) can result in a reduction in particulate emis-sions.

Some fuel quality improvements reduce emissionsfrom all vehicles immediately using the improvedfuel. For example, discontinuing the addition of leadto gasoline instantly eliminates lead emissions fromall gasoline-fueled vehicles, regardless of their age or

12Chapter 3, Cleaner Fuels, Maintaining Fuel Standards; andannexes 1–4.

EXECUTIVE SUMMARY xxiii

state of repair. Similarly, reducing fuel sulfur levelslowers the emissions of oxides of sulfur (SOx). Othermeasures, such as the use of ultralow13 sulfur and theso-called sulfur-free14 gasoline and diesel fuels, areslated to be mandated in industrial countries and afew developing countries during the latter half of this

13In this report, ultralow refers to 50 parts per million (ppm)or lower.

14“Sulfur-free” fuels contain a maximum of 10 ppm sulfur. EUmember states are required to make sulfur-free gasoline and dieselbeginning January 2005, and sell only sulfur-free gasoline and die-sel effective January 2009.

decade primarily to enable the use of sulfur-intolerantexhaust control devices that can dramatically reduceemissions of pollutants, especially particulate matterand oxides of nitrogen (NOx). Because they are a sys-tem, it is important to consider fuel quality and ve-hicular emission standards together. Aside from sig-nificant investment in oil refining capacity that istypically required to produce fuels with very low sul-fur levels, there is also a greater need than is the typi-cal practice in developing countries for proper vehiclemaintenance and operation in order for the exhaustcontrol devices to be effective.

General Guidelines for Fuel Quality

The appropriate standards for fuel will depend on

country circumstances, including the level of air pollu-

tion and the costs of upgrading. But some general

guidelines can be stated:

� Moving to unleaded gasoline as a first priority

while ensuring that benzene and total aromatics

do not rise to unacceptable levels.

� Progressively implementing steps to reduce the

sulfur content of both gasoline and diesel fuels to

very low levels, taking into account the initial situ-

ation and human and financial resource constraints:

– Where the sulfur content of gasoline is high, re-

ducing it to 500 parts per million (ppm) and

preferably lower as soon as possible, to ensure

efficient operation of catalytic converters (fol-

lowing lead removal).

– Where sulfur content in diesel is very high,

identifying and implementing a strategy to re-

duce it to 500 ppm or lower.

– Where moving to 500 ppm for diesel is very diffi-

cult in the near term but lowering it to 2,000–

3,000 ppm is relatively inexpensive, immedi-

ately moving to this level.

– In countries with current or potentially high lev-

els of air pollution from mobile sources, espe-

cially those that have already taken steps to-

ward 500 ppm, or where new or significantly

renovated oil refining capacity is being invested

in, examining the cost-effectiveness of moving

to ultralow sulfur standards, taking into account

maintenance capability and the concomitant in-

vestments in the necessary emission control

technologies to exploit lower sulfur fuels.

� Where the resource and infrastructure conditions

for natural gas are favorable and those for clean

diesel technology are much less so, giving con-

sideration to shifting high mileage public transport

fleets from diesel to CNG.

� Taking steps to prevent fuel adulteration and the

smuggling of low-quality fuels from neighboring

countries, and giving consideration to holding fuel

marketers legally responsible for quality of fuels

sold.

xxiv REDUCING AIR POLLUTION FROM URBAN TRANSPORT

Vehicle technology15

Vehicle technology improvements, including emissioncontrol devices such as catalytic converters and ex-haust gas recirculation, are driven to a large extent byemission standards for new vehicles in industrialcountries. A common policy question for developingcountries is where to set standards for new additionsto their vehicle fleet population (either through im-ports or domestic vehicle manufacture). Countries im-

porting fuels and vehicles find it easier to imposetighter standards than do manufacturing countries. Itis also important to balance standards for new ve-hicles with those for in-use vehicles. If standards fornew vehicles are very stringent and those for in-usevehicles lax, the result is that vehicle renewal becomesexpensive and the retirement of old vehicles may bedelayed. Since old and heavily polluting in-use ve-hicles tend to contribute disproportionately to air pol-lution from mobile sources, it is thus important froman air quality perspective to focus on tightening stan-dards for in-use vehicles and to get them repaired orretired.

General Guidelines for Vehicle Technology

The setting of appropriate vehicle standards

complemented by adequate fuel standards is very

important for emission reduction over time. General

guidelines for setting these standards include the

following:

� Setting emission standards for in-use vehicles at

levels that are achievable by a majority of ve-

hicles with good maintenance, and tightened over

time.

� Progressively tightening vehicle emission stan-

dards for new vehicles to levels consistent with

improving fuel quality.

� Setting emission standards that require the instal-

lation and continued maintenance of catalytic

converters for all new gasoline-powered vehicles

in countries where lead in gasoline has been

eliminated.

� Giving consideration to the introduction of par-

ticulate filters (traps) and other devices to reduce

end-of-pipe emissions from diesel vehicles where

ultralow sulfur diesel is available. As trap and

other device technology develops and prices fall,

and as ultralow sulfur fuels become more widely

available, this strategy will become more robust.

� Establishing regulatory measures to prevent the

import of grossly polluting vehicles.

� Establishing institutions for administering and en-

forcing vehicle emission standards with a primary

task of identifying and removing gross polluting

vehicles from the road.

15Chapter 3, Vehicle Technology; chapter 4, Improving FuelEfficiency through Vehicle Technology; and annexes 2 and 3.

Alternative fuels16

Alternative transport fuels are those other than gaso-line and diesel, and include gaseous fuels, biofuels,and electricity. Although they can be more expensivefor the final consumer than conventional fuels, alter-native fuels can reduce emissions significantly, espe-cially when gaseous fuels replace conventional diesel.Other advantages of alternative fuels include diversi-

fication of energy sources and, particularly in the caseof biofuels, reductions of lifecycle greenhouse gasemissions. The factors needed for successful conver-sion to gas in developing countries include the exist-ence of a gas distribution pipeline for other users ofnatural gas in the case of CNG, close proximity to thesupply of natural gas or liquefied petroleum gas(LPG), and inter-fuel taxation policy that eliminates orreduces the potential financial burden of the substitu-tion of gas for diesel fuel to acceptable limits. Because16Chapter 3, Alternative Fuels; and annex 6.

EXECUTIVE SUMMARY xxv

diesel is taxed much less than gasoline in many devel-oping countries, it is often difficult to stimulate substi-tution of diesel with a gaseous fuel through tax policyalone. When used in conventional vehicles, biofuelssuch as ethanol and biodiesel can have emission ben-efits compared to conventional fuels. The main barrierto biofuels development has been their higher pro-duction costs compared to conventional petroleumfuels, which has meant that, to date, all biofuels pro-grams worldwide have required significant explicit orimplicit subsidies.

Making technical instruments effective17

Technical solutions cannot be viewed in isolation fromtheir policy and institutional context. Where there areserious sector distortions, it becomes much more diffi-cult to implement technical measures. Protection of adomestic auto or oil industry—that is otherwise notable to withstand competition from the internationalmarket—tends to cause technologies and standards tolag. If subsidies are given to the industry or to itsproducts (for example, fuel subsidies), it becomeseven more difficult to tighten standards. An exampleis a state oil company selling subsidized fuels thatfinds it difficult to take even the first step of movingto unleaded gasoline.

Institutional capacity, both public and private, isalso a necessary component for the successful intro-duction of new vehicles, fuels, and emission controltechnologies. Industrial countries have spent decadesputting in place the necessary technical expertise andinfrastructure for proper and regularly conducted ve-hicle maintenance, both in the public and private sec-tors. While developing countries can shorten the timeframe, significant time and resources will be needed

to establish effective institutions for inspection andmarkets for vehicle repair, without which the benefitsof advanced technologies will be greatly reduced.

ConclusionsThere is no simple or universal strategy for reducingtransport sector emissions. While the specific actionsfor reducing transport emissions will vary from city tocity, there are several underlying principles that thisreport seeks to emphasize for building an effectivepolicy package:

� Raise awareness among policymakers and thegeneral public about urban air pollution levelsand damages and specify and promote the rolethat transport plays.

� Press for sector reform that increases sector effi-ciency, benefits society at large by providinggoods and services at lower cost, and at thesame time reduces emissions.

� Raise awareness in business and with consum-ers about business “best practice” that is alsolikely to bring about environmental benefits tosociety.

� Work with, not against, the economic incentivesof various transport actors.

The most aggressive and bold actions to controltransport-related emissions should be undertaken inthose cities with the most serious air quality problemsand where the transport sector is a major contributor.Given the increase in transport emissions that has ac-companied economic growth in virtually every mu-nicipality worldwide, it is important for all cities tobegin putting in place systems for monitoring andcontrolling emissions from the transport sector. How-ever, even where the transport sector’s contribution isnot currently high, such as in major coal-consumingcountries or in low-income cities with a high percent-age of solid fuel use, many of the measures outlinedabove can be appropriate where they have other so-cial and economic benefits.

17Chapter 3, Cleaner Fuels, Downstream petroleum sector re-form; chapter 4, Regulation and Control of Public Road PassengerTransport, Improving internal efficiency of operations; chapter 4,Regulation and Control of Public Road Passenger Transport, Pub-lic transport franchising; and chapter 7.

1

The Context of the Problem

C H A P T E R1

The Air Quality Problem inDeveloping CountriesThe costs of air pollution in terms of damage to hu-man health, vegetation, and buildings and of reducedvisibility are perceived as a serious problem in manydeveloping country cities. Air quality is poorer thanthat in most industrial country cities of equivalentsize. The adverse effects of air pollution often fall dis-proportionately on the poor, compounding the im-pacts from other environmental problems in develop-ing countries such as the lack of clean water andsanitation.

Transport as a Source of PollutionTransport is a known source of many air pollutants.The first six listed below are termed “classic” air pol-lutants by the World Health Organization (WHO):

� Lead from the combustion of leaded gasoline isthe best-known toxin in this context. High leadconcentration in the bloodstream may increaseincidence of miscarriages, impair renal function,and increase blood pressure. Most significantly,lead retards the intellectual development of chil-dren and adversely affects their behavior. Theseeffects may occur even at levels previously con-sidered safe. More lead is absorbed when di-etary calcium or iron intake is low, when thestomach is empty, and when one is very young;for these reasons, poor, malnourished childrenare particularly susceptible to lead poisoning.

� Total suspended particles (TSP), also referredto as suspended particulate matter (SPM), is nota single pollutant, but rather a mixture of manysubclasses of pollutants that occur in both solidand liquid forms. Each subclass contains manydifferent chemical species. Particulate matter

(PM) may be classified as primary or secondary.Primary particles are emitted directly by emis-sion sources, whereas secondary particles areformed through the atmospheric reaction ofgases, such as the reactions between ammoniaand oxides of nitrogen or sulfur that lead to theformation of particles. TSP have historicallybeen monitored and continue to be measured indeveloping countries. The size distribution ofairborne particles matters for health impact. TheWHO places special emphasis on suspendedparticles smaller than 10 microns (µm) in diam-eter (PM10), also called inhalable particulate mat-ter, and those smaller than 2.5 µm (PM2.5), calledfine or respirable particulate matter. Emerging sci-entific evidence points to increasing damagewith decreasing particle diameter. Particleslarger than about 10 µm are deposited almostexclusively in the nose and throat, whereas par-ticles smaller than 1 µm are able to reach thelower regions of the lungs. The intermediatesize range gets deposited in between these twoextremes of the respiratory tract. A statisticallysignificant association has been found betweenadverse health effects and ambient PM10 concen-trations, and recent studies using PM2.5 datahave shown an even stronger association be-tween health outcomes and particles in this sizerange. In response, industrial countries haveswitched from monitoring TSP, which is not di-rectly correlated with health effects, to PM10, andincreasingly to PM2.5.

� Ozone (O3) has been associated with transienteffects on the human respiratory system, espe-cially decreased pulmonary function in indi-viduals taking light-to-heavy exercise. Severalrecent studies have linked ozone to premature

2 REDUCING AIR POLLUTION FROM URBAN TRANSPORT

mortality.1 Ozone also reduces visibility, dam-ages vegetation, and contributes to photochemi-cal smog. Oxides of nitrogen (NOx) and volatileorganic compounds (VOCs) that are photo-chemically reactive (such as aromatics with twoor more alkyl groups and olefins) are the twomain precursors of ozone. NOx is emitted bygasoline- and diesel-powered vehicles, whileVOCs are emitted in most significant quantitiesby gasoline-fueled vehicles. Ambient concentra-tions of ozone are not necessarily lowered by re-ducing the two main contributors to ozone for-mation, hydrocarbons and NOx. Depending onthe ratio of these two components, decreasingone may even increase ambient ozone concen-trations. Therefore, it is important to understandif the atmospheric chemistry in the city falls inthe so-called NOx-limited category (where re-ducing VOC concentrations may have little oreven adverse impact on ambient ozone concen-trations) or VOC-limited category (where reduc-ing NOx may have little or adverse impact onambient ozone concentrations). Mexico City, inpart by virtue of its altitude and geography, hashistorically suffered from high ambient concen-trations of ozone, with levels exceeding theozone standards on about 80 percent of the daysa year. The government has been imposing in-creasingly stringent emission standards on its 3million gasoline-fueled vehicles aided by astrictly enforced emissions inspection program(see annex 7).

� Carbon monoxide (CO), the largest contribu-tors to which are typically gasoline-fueled ve-hicles, inhibits the capacity of blood to carryoxygen to organs and tissues. People withchronic heart disease may experience chestpains when CO levels are high. At very highlevels CO impairs vision and manual dexterity,and can cause death.

� Sulfur dioxide (SO2), which is emitted in directproportion to the amount of sulfur in fuel,causes changes in lung function in persons withasthma and exacerbates respiratory symptomsin sensitive individuals. Through a series ofchemical reactions, SO2 can be transformed tosulfuric acid, which contributes to acid rain andto the formation of secondary (sulfate-based)particulate matter.

� Nitrogen dioxide (NO2) also causes changes inlung function in asthmatics. Like SO2, NO2 canreact to form nitric acid and thereby contributeto acid rain and secondary (nitrate-based) par-ticulate formation. In addition, nitric oxide (NO)and NO2, or NOx as they are commonly called,are precursors of ground-level ozone. Both die-sel- and gasoline-fueled vehicles contribute toNOx emissions.

� Other air toxin emissions of primary concern invehicle exhaust include benzene and poly-aro-matic hydrocarbons (PAHs), both well-knowncarcinogens. Air toxin emissions such as ben-zene depend mostly on fuel composition andcatalyst performance. Exhaust PAHs are dueprimarily to the presence of PAHs in the fuel it-self; in the case of gasoline, they are also formedby fuel combustion in the engine.

Aside from fuel quality, the amounts of pollutantsemitted depend on such factors as the air-to-fuel ratio,engine speed, engine load, operating temperatures,whether the vehicle is equipped with a catalytic con-verter, and the condition of the catalyst.

Air Pollution Levels and TrendsWhile all of these emissions are potentially damaging,their incidence and health impacts differ substantially,both among pollutants and from region to region. Incities with serious air pollution, the three most dam-aging pollutants tend to be lead, particulate matter,and ozone. The strongest evidence linking air pollu-tion to health outcomes is the impact of small par-ticles on premature mortality and morbidity. The con-sistent findings across a wide array of cities, includingsome in developing countries with diverse popula-

1Recent studies have found a significant independent effecton premature mortality for ozone if the important nonlinear ef-fects of temperature (and humidity) are taken into account(Thurston and Ito 1990).

CHAPTER 1. THE CONTEXT OF THE PROBLEM 3

tions and possibly diverse particle characteristics, in-dicate that this association is robust. Particles are evenmore damaging if they contain lead. Ozone is a grow-ing problem in developing country cities and a seri-ous problem in many industrial countries.

WHO studies of megacities show that, althoughhealth-based guidelines of pollutants can be widelyexceeded, the significance of the problem varies con-siderably. Lead excesses over norm are serious whereleaded gasoline is used, but not usually elsewhere.Excess of CO is typically not nearly as great as that ofsmall particulate matter, especially in countries wherethe consumption of gasoline is relatively low com-pared with that of diesel.2 Significantly elevated levelsof ambient SO2 tend to come from the combustion ofcoal much more than from the transport sector. Ma-rine engines may also contribute disproportionatelyto the local SO2 inventory in port cities because ma-rine fuel tends to be high in sulfur.

However, the situation is not static. Ambient NO2

concentrations are often below the WHO guidelinesbut are on the increase, as are those of ozone. As inthe industrial countries, it is expected that the relativeimportance of mobile source air pollution will in-crease in developing countries as incomes grow andother gross polluters either disappear naturally (do-mestic wood burning), or are suppressed at source(industrial pollution).

In designing measures to improve air quality, it istherefore necessary to take into consideration both thecurrent pollutants of concern and likely future trends.If ozone is within air quality limits but on the rise andthe gasoline vehicle population is growing rapidlydue to rising household income, ozone is likely to be aserious problem in the future unless steps are taken.The suitability of different measures should also takeinto account the climatic conditions, topography ofthe city, altitude, dispersion profiles, and other defin-ing characteristics of the airshed, all of which affectambient pollutant concentrations in addition tosources and levels of emissions.

Global Climate ChangeTransport is a growing source of greenhouse gas(GHG) emissions, which are believed to be contribut-ing to a change in the earth’s climate. Major GHGemissions from mobile sources are carbon dioxide(CO2), methane, nitrous oxide (N2O), and chlorofluo-rocarbons (CFCs) or other refrigerants. Over a 100-year horizon, the global warming potential of meth-ane and N2O is estimated to be 23 and 296 times thatof CO2, respectively, on a weight basis. The potentialof CFCs is thousands of times greater than that ofCO2. Recently, soot has been receiving increasing at-tention as a contributor to global warming. Some sci-entists say soot, such as from diesel engines, is caus-ing as much as a quarter of all observed globalwarming by reducing the ability of snow and ice toreflect sunlight (Hansen and Nazarenko 2004). Oiland natural gas production can release a largeamount of GHGs: one oil company reported that itburned one barrel of oil in its entire operation for ev-ery three extracted from its wells (Automotive Envi-ronment Analyst 2001). Flaring of associated naturalgas is still common, releasing CO2 and residual un-burned methane. Worse still is the venting of naturalgas, released in the form of methane, a much morepowerful GHG than CO2. Complete combustion ofcarbon-containing fuel in a vehicle produces CO2,while N2O has been found in recent years to beformed in significant quantities in vehicles equippedwith aged three-way catalytic converters. Leaks of re-frigerants used in air conditioning are another sourceof GHG emissions. Vehicles fueled by natural gastend to have higher methane emissions.

In order to properly assess the contribution of par-ticular transport fuels to GHG emissions, a full life-cycle analysis (not just an analysis of tailpipe emis-sions) is needed. That analysis should includeemissions that occur during the preparation of thefuel supply and vehicle manufacture. Refining pro-cesses are energy-intensive, and energy intensity in-creases with increasing stringency in fuel specifica-tions, especially with respect to sulfur reduction.3 A

2Ambient CO concentrations can be high at certain “hotspots” such as traffic corridors and intersections.

3For example, to reduce sulfur levels in fuels in the refiningprocess, a source of hydrogen is needed to bond with the sulfur


recent well-to-wheel analysis, jointly carried out byauto and oil companies in conjunction with theArgonne National Laboratory in the North Americancontext, examined advanced fuel and vehicle systemsthat had reasonable chances of being commercializedin large volume—oil-based fuels, natural gas-basedfuels, hybrid, hydrogen, and alcohols (General Motorsand others 2001). The study showed that vehicles fu-eled by woody or herbaceous cellulose-based ethanol(as opposed to ethanol based on agricultural cropssuch as corn) had by far the lowest GHG emissions.This was followed by hydrogen fuel cell hybrid elec-tric vehicles. Diesel hybrid electric vehicles had GHGemissions that were significantly lower than those ofconventional gasoline vehicles.

Controlling CO2 emissions from the transport sec-tor has been found to be generally more difficult thancontrolling emissions from other sectors in industrialcountries. A recent study by the Organisation for Eco-nomic Co-operation and Development (OECD) Work-ing Group on Transport focused on policy instru-ments and strategies for achieving environmentallysustainable transport in Canada and eight Europeancountries (Working Party on National EnvironmentalPolicy 2002). Because of the difficulty of stabilizing,let alone reducing, transport sector emissions, thecontribution of the transport sector to total CO2 emis-sions in OECD countries is forecast to increase fromapproximately 20 percent in 1997 to 30 percent in2020 (Environment Directorate Environment PolicyCommittee 2002). In developing countries, with risingincome and the rapidly rising mobilization that ac-companies it, the increase in CO2 emissions in thecoming years will be even greater than in OECDcountries in absolute tonnage.

Even more so than for local air pollutants, it is es-sential to identify interventions with multiple socialand economic benefits for controlling GHG emissionsin developing countries. According to the Science andTechnology Advisory Panel (STAP) of the Global En-vironment Facility (GEF), measures that affect the

long-term energy demand by the urban transport sec-tor, such as modal choice and land use planning, arelikely to have much greater effect on GHG emissionsand be more cost-effective than incremental changesin fuels and vehicles (GEF 2002).

Urban Transport Policy in DevelopingCountriesEnvironmental policy decisions cannot be separatedfrom transport sector policy decisions. Urban air pol-lution from mobile sources is a by-product of the pro-duction of urban transport services. Those servicesare essential to the economic health of a city and tothe welfare of all its inhabitants, including the poor.Environmental and ecological impacts are important,but are only one aspect of urban transport policy; eco-nomic, financial, social, and distributional concernsalso come into play. The various dimensions of policyimpact have to be balanced in the local and nationalpolitical process.

The three dimensions of policy concern—those re-lating to economics, the environment, and equity—may sometimes be in harmony. For example, the costof eliminating lead from gasoline is usually small incomparison to the large health benefits that typicallyaccrue to some very vulnerable groups, particularlypoor, undernourished children. Other urban air qual-ity issues are not so straightforward. For example, theimposition of stringent fuel and vehicle emission stan-dards can increase capital and operating costs for busoperators or transit systems. Where there are not com-pensating reductions in operating costs, fares mayhave to be raised, which in turn might make the ser-vice unaffordable by poor users. Transport policymust be designed to be both environmentally sensitiveand consistent with public and private affordability.

Both on the supply and the demand sides, the ur-ban transport sector is very fragmented. Moreover,most transport actors—individual and corporate us-ers of transport services as well as transport suppli-ers—are motivated by private benefit or profit. Be-cause air pollution impacts are typically external tothe transport actors, and uncharged for in financialterms, they tend to be discounted or ignored in the

for removal. The higher the sulfur level of the crude oil, the morehydrogen and energy will be needed for sulfur reduction of re-fined products such as gasoline, diesel, and fuel oil.

CHAPTER 1. THE CONTEXT OF THE PROBLEM 5

private decisionmaking process of transport users orsupply agencies. Whenever possible, therefore, strate-gies to reduce air pollution from mobile sourcesshould be designed to be seen as in the private inter-ests of the actors as well as in the social interest. Forthese reasons, fiscal measures are an essential supportfor environmental initiatives. Technological improve-ments and physical measures may therefore be seennot as substitutes for, but as complements to, fiscalmeasures such as taxes, penalties, and subsidies.

The Policy StanceFrom all these considerations it may be expected thatwhile the environmental goals will be largely sharedamong countries, the means of achieving these goalsamong different countries and cities within countrieswill vary according to their level of economic activity,air quality problems, and climatic and geographicalconditions. Relatively wealthy developing countrycities are likely to face problems and seek solutionssimilar to cities in industrial countries where theadoption of modern technologies has contributed sig-nificantly to the reduction of urban air pollution. Invery poor countries with limited levels of motoriza-tion, there may be other environmental and social

problems that are more pressing than urban air pollu-tion. While state-of-the-art transport technologies willalso play a role in limiting air pollution in these coun-tries, resource limitations and competing prioritiesmay make a different time scale of introduction ap-propriate.

To be effective and sustainable over the long term,regulatory and policy instruments for reducing trans-port emissions must provide incentives for individu-als and firms to limit the pollution from existing ve-hicles and to avoid delay in adopting new and cleanertechnologies and fuels. Public and private institutionsmust be equipped with the resources and the skillsnecessary to monitor transport emissions and takemeasures to reduce them. Above all, measures mustbe cost-effective and affordable in light of the myriadof other pressing needs in developing country cities.

The aim of this report is not to provide a “one-size-fits-all” prescription of what all countries shoulddo immediately. Instead, it is intended to provide in-formation and advice on air pollution control experi-ences in the transport sector from both industrial anddeveloping countries that will assist the local formu-lation of policies towards urban air pollution in manydifferent circumstances.

7

A Systematic Approach to ControllingUrban Air Pollution from Mobile Sources

To assess the seriousness of transport-related air pol-lution and enable informed decisionmaking to com-bat air pollution, it is useful to have a systematicframework that can guide the process to evaluate is-sues and available data and to arrive at workable so-lutions. This chapter describes the type of questions topose and methodologies that may be pursued in ad-dressing mobile sources of air pollution.

A Framework for AnalysisGiven the diversity of problems and situations, it isimportant to ask the right set of questions in order todiagnose urban air pollution problems, determine therole of the transport sector, and identify affordableand sustainable solutions. Figure 1 suggests a numberof logical steps in developing a strategy for control-ling air pollution from urban transport.

The first step is to establish the magnitude and na-ture of the ambient air quality problem in a particular

Sequence of Questions to Appraise Mitigation Options to Tackle Mobile SourcesF I G U R E

1

Yes

No

Yes

No

Priority: Are the adverse impacts

of outdoor air pollution serious?

Problem pollutants: Whichpollutants cause

the most damage?

Policy alternatives: Whichpolicy instruments will minimize distortions and achieve results

cost-effectively?

Sources within transport: Which transport activities

do the most damage?

Which environmental benefits can be achieved through no regret adjustments to transport policy?

Role of transport: Does transport

contributesignificantly to the

problem pollutants?

city. The most damaging pollutants (on the basis ofthe combined impact of high ambient concentrations,toxicity, and exposure) then need to be identified.Thereafter it is necessary to determine the relativecontribution of mobile sources (transport), and to es-tablish which transport activities are the largest con-tributors of the most damaging air pollutants. Finallythe ways these activities can be improved must beanalyzed and the instruments by which the improve-ment could be achieved need to be identified. The ef-fectiveness of different policy instruments should beassessed in order to construct the appropriate localpolicy package.

Air Quality Monitoring and StandardsThe level and nature of air pollution varies substan-tially from city to city. Hence, the first requirement isthe creation of an adequate knowledge base on localair quality on which to develop an air quality policy.

C H A P T E R2


Monitoring of ambient air quality is an important firststep. It is especially important to measure small par-ticulate matter. In a number of developing country cit-ies, ambient concentrations of PM10 or PM2.5 are notmeasured regularly, or not measured at all. The ab-sence of ambient data on what is probably the mostimportant pollutant of concern makes it difficult toquantify the seriousness of outdoor air pollution.Common obstacles to systematic monitoring of PM10

and PM2.5 are a shortage of skilled staff and of fundsto operate, maintain, and repair instruments. Forozone, oxides of sulfur (SOx), NOx, and CO, rapid as-sessment can be carried out using diffusion tubes,which are relatively inexpensive to deploy and aregood for giving spatial distributions of pollutant con-centrations (averaged over several days) when theyexist at elevated levels. Seasonal variation also needsto be established. Quality assurance and quality con-trol should be given adequate attention to ensure datareliability.

The measured concentrations can be compared tonational air quality standards to see where standardsare exceeded. Another useful guide is WHO’s health-based air quality guidelines, which provide numericallimits for different averaging periods for all the classicpollutants except particulate matter (WHO 2000). Inthe case of PM10 and PM2.5, WHO gives no numericallimits on the grounds that no threshold levels for mor-bidity and mortality exist. In their absence, many de-veloping countries follow the U.S. or European stan-dards for particulate matter.

In a number of developing country cities, ambientconcentrations of PM10 are grossly in excess of the na-tional air quality standards while other pollutant con-centrations seldom exceed them. In a few higher-in-come developing country cities, ozone has become aserious problem, and this problem is on the increase.One of the best-known examples is Mexico City,where ozone exceedances are much more commonthan PM10 exceedances. In cities with extensive use ofcoal, PM10 and SO2 are the main pollutants exceedingthe national air quality standards by a large margin.

Air pollution becomes a problem of policy concernwhen ambient concentrations of harmful pollutants

are elevated and when a large number of people areexposed. The latter—significant human exposure—isthe rationale for relocating polluting industries out-side of cities or supplying cleaner fuels to powerplants near major cities. Because human exposure isan important factor, air quality monitoring should tar-get large population centers first. Within a metropoli-tan area, monitoring should take place at a variety oflocations: urban “hot spots” such as areas affected byvehicle or industrial emissions; city center pedestrianprecincts, shopping areas, and residential areas repre-sentative of population exposure; and urban locationsdistanced from sources and therefore broadly repre-sentative of city-wide background conditions, such aselevated locations, parks, and urban residential areas.

The Determinants of Transport Emissions

Transport activity characteristicsThe major source of ambient lead is combustion ofleaded gasoline: ambient lead concentrations have al-ways been reduced significantly when lead has beenbanned in gasoline. However, while many countrieshave now banned lead, use of leaded gasoline contin-ues in some countries, including Indonesia, Venezu-ela, and a number of countries in Sub-Saharan Africa.It is especially problematic in two-stroke engines be-cause some alkyl lead added to gasoline as an octaneenhancer may be emitted uncombusted as organiclead, which is even more damaging to health than theinorganic lead formed as a result of combustion ofalkyl lead.

Fine particles are emitted as a product of combus-tion, especially from diesel vehicles but even fromnatural gas-fueled vehicles as a result of combustionof lubricants. Particulate emissions can increase sub-stantially where engines are underpowered or poorlymaintained or adjusted. Black diesel smoke resultsfrom inadequate mixing of air and fuel in the cylinder,with locally over-rich zones in the combustion cham-ber caused by higher fuel injection rates, dirty injec-tors, and injection nozzle tip wear. Overfueling to in-crease power output, a common phenomenonworldwide, results in higher smoke emissions and

CHAPTER 2. A SYSTEMATIC APPROACH TO CONTROLLING URBAN AIR POLLUTION FROM MOBILE SOURCES 9

somewhat lower fuel economy. Dirty injectors arecommon because injector maintenance is costly interms of actual repair costs and of losses stemmingfrom downtime. Adulteration with heavier fuels alsoincreases in-cylinder deposits and fouls injectors.While most modern gasoline vehicles return thecrankcase emissions to the engine intake and re-burnthem in the cylinder, all diesel engines still vent to theatmosphere because, although it can be done, thereare technology barriers to re-circulating the crankcasegases. These crankcase emissions consist mainly ofunburned diesel, products of combustion, partiallyburned diesel, and products from the lubricating oil(oil mist and some products of thermal degradation ofoil).

Particulate emissions from gasoline vehicles tendto be much lower in mass than those from diesel ve-hicles,1 except in the case of two-stroke engine gaso-line vehicles where “scavenging losses” and the directintroduction of lubricant along with the intake air-fuelmixture lead to high emission of “white smoke.”White smoke mostly comprises fine oil mist andsoluble hydrocarbons, whereas the black smoke emit-ted by diesel vehicles contains a large fraction of gra-phitic carbon. The health impact of white smoke is notwell understood.

Non-exhaust particles can also contribute signifi-cantly to overall particulate emissions from the trans-port sector. The factors affecting the total level of non-exhaust particulate emissions include:

� Tires and their interaction with different roadsurfaces

� Brake pad/shoe dust� The operating characteristics of vehicles (for ex-

ample, speed, acceleration, and loading)� Type of road (paved versus unpaved)� Ambient weather conditions (for example, tem-

perature, rain, and wind).

The health impact of some categories of non-ex-haust particles is believed to be serious because they

1Emerging evidence suggests that both gasoline and diesel ve-hicles emit “nanoparticles.” The perception of gasoline particulateemissions may change if mass is no longer the metric for particles.

are typically very small, with an average aerody-namic diameter of just 1 micron for bitumen particles.There is growing evidence of a relationship betweenthe incidence of asthma and the concentrations ofnatural latex protein particles in the atmosphere re-sulting from the abrasion of tires and roads. Tires usea blend of natural rubber (latex and dry sheet) andsynthetic rubber. Since the early days of paved high-way construction, rubber has also been added tomodify asphalt. Its release following abrasion withvehicle tires increases the concentration of rubber par-ticles in the atmosphere. Unpaved roads create a dif-ferent problem: they are responsible for significant re-suspension of road dust. It is difficult to quantify theimpact of non-exhaust particles on overall ambientconcentrations in developing countries. Most of thedata collected so far are from high-income industrialcountries. This area merits greater attention.

Gasoline vehicles are the dominant contributor toambient concentrations of CO in most cities. They arealso responsible for emissions of NOx and photo-chemically reactive VOCs which, in the presence ofsunlight, react together to form ozone. Secondary par-ticles are formed from NOx and SOx, which are alsoresponsible for acid rain with adverse effects on veg-etation. In addition, sulfur acts as a poison for cata-lysts, although its adverse effects are partially revers-ible in conventional three-way catalysts. SOx

Particulate emissions, some of which exit the tailpipe as black

smoke, increase significantly when vehicles are in poor state of re-

pair or engines are improperly adjusted, such as this heavy-duty die-

sel-powered bus.


emissions are directly proportional to the fuel sulfurcontent. While gasoline sulfur levels tend to be low, asmuch as 1 percent (10,000 parts per million [ppm])sulfur is still allowed in automotive diesel in some de-veloping countries.

The transport operating characteristics determin-ing the sector’s contribution to urban air pollution arecomplex. For any given vehicle and fuel combination,aggregate emission levels vary according to the dis-tance traveled and the driving pattern. The emissionsof CO2 and SOx vary directly with fuel consumption.The tailpipe emissions of CO, NOx, particulate matter,and hydrocarbons vary in addition with the air-to-fuel ratio, injection timing, and other settings. Figure 2gives an illustration of particulate emissions from die-sel vehicles as a function of vehicle speed. The opti-mum steady speed for emissions, which is usually inexcess of 60 kilometers per hour (km/h), is rarelyachievable in urban areas. Broadly speaking, engine-out NOx emissions increase, and CO, particulate, andhydrocarbon emissions decrease, with increasing en-gine temperature or increasing vehicle speed.

The driving cycle has a significant influence onemission levels for a given vehicle; for example, bothfuel consumption and pollutant emissions are manytimes higher per vehicle km during acceleration anddeceleration than during cruise. Cold-start enrich-ment of gasoline engines increases CO emissions or-ders of magnitude more than in warm operation.

Particulate Emissions as a Function of VehicleSpeed

F I G U R E2

Source: Ntziachristos and Samaras 2000.

0.00.51.01.52.02.53.03.54.0

Vehicle speed (km/h)

g PM/km

0 10 20 30 40 50 60 70

Medium buses

Trucks

Heavy buses

Moreover, as catalytic converters depend on heat fortheir effectiveness, they are least effective at cold start,further accentuating the influence of the driving cycle.

Human exposureCosts to society arising from urban air pollution in-clude damage to buildings and vegetation, loweredvisibility, and heightened GHG emissions. However,increased premature mortality and morbidity are gen-erally considered to be the most serious conse-quences, because of both their human and economicimpacts. It is common and appropriate, therefore, touse damage to human health as the primary indicatorof the seriousness of air pollution. To assess this dam-age it is necessary to understand the toxicity of differ-ent pollutants, the ambient levels above which theybegin to have health impacts (the threshold), and thenumber of people exposed to different levels of airpollution above that threshold.

The effect of any specific airborne pollutant de-pends on the location of the emissions with respect tothe human population and the ways and extent towhich emissions are dispersed. For example, evenpollution from gross emitters may not be importanton deserted roads where virtually no human beingsare exposed, or in very windy locations from whencethe emissions are rapidly dispersed. In practice, be-

cause urban vehicle emissions are emitted near

ground level where people live and work, urban pol-

lution from mobile sources merits the attention of

policymakers in many cities.

City size and incomeTransport-related air pollution tends to be worst inlarge and densely populated cities, particularly thosethat are highly motorized, where there are high levelsof congestion, and where there is a proliferation ofolder, more polluting vehicles.

For the highest-income cities in developing coun-tries, as well as for cities in some of the transitioneconomies of Eastern Europe that had already begunto develop car ownership before the transition, levelsof motorization are high and equipment is sophisti-cated. In such cities the problems are likely to be simi-


lar to those in industrial countries where ozone is im-portant, in addition to particulate matter.

Assessing Air Pollution MitigationMeasures

The “opportunity cost” approachAll commitments of public resources are implicitlyeconomic decisions because they involve the diver-sion of resources from alternative uses. The goal of acost-benefit analysis is to compare the monetized ben-efits of a policy—for example, a policy to reduce airpollution—with its costs. The value of the resourcesused in the best alternative use is called the “opportu-nity cost.” The difference between benefits and costsis termed the net benefit of the policy. A cost-benefitanalysis may be performed to determine whether agiven policy yields positive net benefits, or to set localambient air standards, or to help policymakers rankvarious pollution control policies.2

In the case of an air pollution control policy, thefirst step in the analysis of benefits is to predict emis-sions of the common air pollutants for all major pol-luting sectors, with and without the policy. The sec-ond step in the calculation of benefits is to translatethe emissions predictions into ambient pollution con-centrations with and without the policy. This allowsthe analyst to calculate the change in ambient PM orozone concentrations attributable to the policy.

The damage function approach is then used toquantify the physical benefits associated with thepolicy and to value them. This entails translatingchanges in ambient concentrations into physical ef-fects—for example, calculating the cases of illness andpremature mortality avoided by the policy—and thenvaluing these physical effects. A similar approach isused to estimate the agricultural and aesthetic ben-efits associated with reducing air pollution. After ben-efits have been calculated, they can be compared withthe costs of the policy.

Attributing ambient pollution to specific sourcesOne of the most difficult and costly steps in the evalu-ation of an air quality regulation is to translate emis-sions with and without the regulation into ambientpollution concentrations. This is usually accom-plished with air quality models that are demandingboth in terms of the inputs they require (detailed me-teorological and atmospheric data, in addition toemissions inventories) and in terms of the computertime they use (for example, four days of computertime to simulate a seven-day air pollution episode). Inpractice, the state of air quality modeling limits thenumber of regulatory options that can be evaluated ina cost-benefit analysis. This implies that it is impor-tant to screen possible regulatory options before per-forming a full-blown cost-benefit analysis.

Though the impacts of urban air pollution havebeen documented in both industrial and developingcountries, a fundamental policy question is to whatextent ambient air pollution is contributed by mobilesources (cars, trucks, buses, motorcycles) as opposedto stationary sources (power plants, industry, com-mercial establishments, households). Even more sothan for stationary sources, there are multiple actorsin the transport and transport fuel-supply sectors thatmust be made part of the strategy for reducing mo-bile-source emissions.

While measurements of ambient air quality pro-vide information on the level of pollution, they saynothing directly about its sources. Broadly speaking,there are two approaches to quantifying the contribu-tions of pollution sources to ambient concentrations:

� Dispersion modeling starts with the estimatedemissions from different sources (called the“emissions inventory”) and, on the basis of amodel of how those emissions are dispersed,calculates the expected ambient concentrationsat particular “receptor” sites where ambientconcentrations are measured. Ambient concen-trations can be used to calibrate the dispersionmodels for running future scenarios. Each pol-lutant considered a significant hazard should bemodeled in this way. An example of this ap-proach in developing countries can be found in

2Policies are ranked according to the size of their net benefits(benefits minus costs) or, when budgets are limited, by the ratio ofbenefits to costs.


the Urban Air Quality Management Strategy inAsia (URBAIR) reports (Grønskei 1996a and1996b, Larssen 1996a and 1996b). Modeling ofsecondary particulate formation as well as con-tribution of distant sources adds significantcomplexity.

� Receptor modeling uses detailed chemicalanalysis of particles in the atmosphere to matchtheir characteristics at given receptor and sourcelocations (“fingerprinting”). Unfortunately, it israre for one compound or element to be exclu-sive to a single source and hence to act as an un-ambiguous tracer for that source. More com-monly, similar sources may have dissimilarprofiles, while different source categories mayhave similar profiles. Recently, a PM2.5 sourceapportionment study using receptor modelingwas conducted in three Indian cities (ESMAP2004). Detailed speciation of particles conductedin properly executed studies is time-consumingand resource-intensive. Nevertheless, even rudi-mentary carbon and other chemical analyses of

particles can give a broad-brush picture of con-tributions from different sources.

In carrying out source apportionment, identifica-tion of distant, contributing, upstream sources is alsoimportant. Over the longer term, significant invest-ment is needed both in equipment and laboratory fa-cilities and in human resources, which can be estab-lished at national or private research institutes oruniversities. Megacities may be able to carry out theseactivities, smaller cities may not. Depending on theextent of pollution damage in smaller cities, it maymake sense to have a centrally funded facility to pro-vide monitoring and analytical skills to smaller cities.

Several important lessons can be learned from ex-perience in the attribution of pollution amongsources.

� What ultimately should drive policy is notwhich source is emitting more, but which sourceis likely to lead to greater exposure to health-damaging pollutants and at what cost thatsource of emissions can be mitigated. A coal-

Source Apportionment: Lessons from the United States

Dispersion modeling and receptor modeling are two main approaches to quantifying the contributions of pollution

sources to ambient concentrations. In principle, both approaches should give the same results. In practice, em-

pirical studies comparing the results of the two approaches are rare, even in industrial country cities.

One of the most extensive comparisons of the two approaches is a study in Colorado (Watson and others 1998),

which examined source contributions to PM2.5

. The available emissions inventory indicated that diesel accounted

for two-thirds of on-road vehicle PM2.5

emissions and gasoline the remaining one-third. However, the use of a re-

ceptor model suggested that diesel actually accounted for only one-third and gasoline two-thirds, and that PM2.5

emissions from gasoline vehicles were seriously underestimated, not only with respect to diesel but also on an

absolute basis. The discrepancy was caused primarily by the presence of gasoline “smokers” and high emissions

during cold start.

A study conducted in southern California (Durbin and others 1999) found that some gasoline-fueled passenger

cars emit as much as 1.5 grams per kilometer (g/km), an emission level normally associated with heavy-duty die-

sel vehicles. Comprising only 1–2 percent of the light-duty vehicle fleet, these gross polluters were estimated to

contribute as much as one-third to the total light-duty particulate emissions. It is possible that the proportion of

“smoking” gasoline vehicles is much larger in developing countries.

Source: World Bank 2002a.

I N T E R N A T I O N A L

E X P E R I E N C E 1


fired power plant with a tall stack that is locatedat the edge of a city may be the largest emitter ofparticles in terms of absolute tonnage, but maybe contributing less—from the point of view ofoverall human exposure—than all the house-holds burning biomass. The height at which pol-lutants are emitted, and where they are emitted,matters a great deal. An emissions inventory,

which typically ranks pollution sources accord-

ing to their absolute primary emissions, should

not be the sole basis of air pollution control

policy since it does not account for the contribu-

tion to ambient concentrations and the exposure

of human populations.

� Total weight of pollutants is also a poor indica-tor of impact. It is often concluded that roadtraffic is by far the largest contributor to urbanair pollution, because in absolute tonnage, COdominates all other pollutants, and the majorityof CO is from gasoline vehicles. But the toxicityof CO is much lower on a weight basis than thetoxicities of other pollutants, so that these resultscannot be directly correlated with health effects.Different pollutants with varying toxicities

should not be added together on a weight basis

as a way of indicating the seriousness or princi-

pal sources of air pollution.

Vehicle exhaust’s contributions to ambient concen-trations of PM10 and PM2.5 in different cities are shownin table 1.

Because of the particle size distribution of vehicu-lar emissions, vehicle exhaust constitutes a larger frac-tion of PM2.5 than of PM10 at any given location. Ve-hicle exhaust contributes little to ambient air inQalabotjha in South Africa and Teplice in the CzechRepublic where it is dominated by coal combustion,as has been the case in the majority of cities in China.The study in Denver, Colorado, is one of the most ex-tensive undertaken to date (see International Experi-ence 1). The study of 17 cities in the United Kingdomlooked at the contribution of road traffic to PM10 inwinter and summer as well as over the entire year.The contribution in winter was consistently higherthan that in summer in every city. The contribution of

vehicle exhaust varies significantly from site to site,ranging from a negligible percentage to more thanhalf of the ambient particulate concentrations.

An accurate vehicle registration system is crucialfor obtaining information on the vehicle populationand fleet characteristics. This, together with estimatesof annual vehicle distance traveled and the total salesof automotive fuels, can indicate which vehicle cat-egories are likely to be major contributors to vehicularemissions. Having a reliable vehicle registration sys-tem is also a prerequisite for an effective vehicle emis-sions inspection system.

Analysis of the impacts of air pollution on healthThe approach most commonly used to value healtheffects of air pollution is known as the “damage func-tion approach.” This involves estimating the impactof a change in level of air pollution on health and thenattributing a monetary value to the change in health.The benefits of urban air quality improvements canthen be evaluated in terms of the particular nationalsituation and resource availability.

The analysis of the health benefits associated withair pollution reduction has made great progress overthe past 10–15 years. Estimates of the health impact ofair pollution are generally obtained from epidemio-logical studies that are designed to determine rela-tionships—referred to as concentration-response (CR)functions—between air pollution and health effects inhuman populations (see annex 8 for more detail).While many questions remain, the accumulation of

studies over the years has yielded considerable con-

sistency on the impacts of specific pollutants, espe-

cially the mortality impact of particulate matter.

Valuing the health effectsTo convert the health impacts into monetary termsthat can be compared with the resource costs of reme-dial policies, economists have customarily placed avalue on avoided morbidity based on the amount thata person is estimated to be willing to pay to avoid theillness (World Bank 2003b). Where estimates of will-ingness to pay (WTP) are not available, it is custom-ary to use the avoided medical costs and productivitylosses arising as a consequence of the illness. This


Contribution of Vehicle Exhaust to Ambient Particulate Concentrations

City PM contribution Technique Source

Mumbai, India 30% of ambient PM10

Dispersion modeling Larssen and others 1996b

Bangkok, Thailand 30% of annual and geographical Dispersion modeling Radian 1998average PM

10 in 1996

Mexico City, Mexico 47% of ambient PM2.5 (daytime) Receptor modeling Vega and others 199740% of ambient PM2.5 (nighttime)in 1989–90

Teplice (industrial town), Average 2.8% of PM2.5

in winter 1993 Receptor modeling Pinto and others 1998Czech Republic

Qalabotjha, South Africa 0.1% of ambient PM2.5

(study period Receptor modeling Englebrecht and(township south of winter 1997) Swanepoel 1998Johannesburg)

Sihwa, Republic of Korea 8.6% of ambient PM2.5 (4 intensive Receptor modeling Park and others 2001(residential area near periods in 1998 and 1 intensiveheavy industrial complex) period in 1999, each of 10 days)

Denver, Colorado, USA 55% of ambient PM2.5

Receptor modeling Watson and others 1998

17 U.K. cities Percentage of annual PM10 in 1996: Receptor modeling Airborne Particles ExpertLondon Kensington 33% Group 1999, table 5.1Lowest contribution of road transport to PM

10—Swansea 17%

Highest contribution—Wolverhampton 52%

T A B L E1

“cost-of-illness” approach, which does not include thevalue of pain and suffering and lost leisure time, isusually a lower bound to the WTP. For prematuremortality, economists have tried to measure whatpeople would be willing to pay to reduce the risk ofdying, usually expressed in terms of the value of astatistical life (VSL). When estimates of the VSL areunavailable, forgone earnings (including adjustmentsfor labor market distortions) are often used to place alower bound on the VSL. In practice it has been diffi-cult to obtain reliable estimates of the VSL, especiallyfor developing countries. This is also true for esti-mates of the WTP to avoid morbidity. Annex 9 de-scribes various options and discusses how these cal-culations may be carried out in developing countries.Despite the difficulties and large uncertainties associ-ated with valuing avoiding morbidity and mortality,it is still advisable to apply the principles of cost-

benefit analysis as a method of prioritizing interven-

tion measures.

Non-health impactsNon-health impacts of urban air pollution includedamage to property (for example, sooting or corro-sion), damage to the local economy resulting from de-cisions to locate activities away from heavily popu-lated areas, and the disutility to residents ofunpleasant, albeit not necessarily health-threatening,conditions. Non-health impacts are usually more dif-ficult to quantify and evaluate than are health im-pacts. What studies of air pollution have shown, how-ever, is that the health impacts generally greatlyexceed non-health impacts in socioeconomic terms.Health benefits may be a reasonably proxy for total

benefits, particularly in comparisons between differ-

ent air pollution instruments.

The results: identifying the most damagingpollutantsWhen risk assessment of susceptibility to physical ex-cesses is combined with evidence of health impacts


from concentration-response (CR) analysis, studies ina number of cities (for example, Bangkok, Cairo,Mexico City, Quito, and Santiago) have indicated thatthe greatest damage to human health comes from ex-posure to small particulate matter. In some citieswhere a large amount of leaded gasoline is still con-sumed, airborne lead is another damaging pollutant.Depending on topographical and meteorological con-ditions, ozone can also be a serious health problem inlarge metropolitan regions, as it is in Mexico City,Santiago, and São Paulo. More detail is given in an-nex 8, including discussion of other pollutants.

Understanding the health impacts of air pollutantshas been developing rapidly. Only within the pasttwo or three decades have the health impacts of par-ticulate matter and lead at even relatively low ambi-ent concentrations attracted serious attention by re-searchers and policymakers. In recent years, thescience of particulate air pollution has increasinglypointed to the importance of the number and size ofparticles (especially of ultrafine particles), rather thanof aggregate mass (World Bank 2003c). The widely ac-cepted notion that gasoline vehicles emit far less par-ticulate matter than conventional diesel vehicles,based on total mass, may change if particle size andnumber come to be included in emission regulations.The Motor Vehicles Emissions Group of the EuropeanCommission is giving consideration to a proposal tolimit the number of particles, in addition to mass, invehicle exhaust.

Particulate emissions from the combustion oftransport fuels and vehicle tire wear fall predomi-nantly in the submicron range, raising serious healthconcerns. As a result, policy formulation has begun tofocus on the contributions of combustion processes tothe size fractions now considered most damaging topublic health. A study in the United Kingdom re-ported that road traffic nationally contributed 25 per-cent of primary PM10 emissions, but the relative im-portance of road traffic emissions increased withdecreasing particle size, and road transport accountedfor an estimated 60 percent of PM0.1 (particles smallerthan 0.1 microns, also called ultrafines) (Airborne Par-ticles Expert Group 1999). There is also a growingconsensus that diesel exhaust poses a serious cancer

risk (Lloyd and Cackette 2001), and diesel exhaust isclassified as a probable human carcinogen by manygovernmental authorities, including the InternationalAgency for Research on Cancer (part of WHO), theU.S. National Toxicology Program, and the U.S. Envi-ronmental Protection Agency (EPA). The impacts ofsome unregulated emissions, such as the so-called air-borne toxics, are not as well understood because it ismore difficult to estimate human exposure. The cur-

rent understanding of pollution impacts shows that

fine particulate matter should be a major focus of

control efforts.

Cost-Effectiveness AnalysisGiven the difficulties of calculating the benefits of re-ducing morbidity and mortality, another option, dis-cussed here, is to compare policies by their cost-effec-tiveness, for example by their cost per unit change inambient pollutant concentrations.

The scope for cost-effectiveness analysesEven if the monetary evaluation of the benefits of im-proved air quality may be elusive, the costs of themeasures to obtain given physical improvements maynot be. An alternative form of economic evaluation isthat of cost-effectiveness analysis, which comparesthe costs of achieving different levels of improvementin physical conditions. Cost-effectiveness analysis

can give strong guidance to decisionmakers on what

are preferable instruments or packages of instruments

to use.

The simplest form of cost-effectiveness analysiscomputes the cost per ton of emissions reduced forvarious regulatory options. Even this approach givesestimates with large uncertainties because of the na-ture and number of assumptions that are necessarilyinvolved. That said, cost-per-ton calculations providea useful way to screen air pollution control policies.Because the height at which particles are emitted isessentially the same for all vehicle types, the impacton ambient air quality of reducing a ton of particulateemissions from different vehicles or of using differenttransport policy options is likely to be similar, pro-vided that the temporal and geographical distributionof vehicle operation is also comparable.


Regulators typically face the problem of choosingamong a set of regulatory options that vary in the sizeof the reduction in air pollution they deliver and intheir expense. The purpose of a cost-effectivenessanalysis of air pollution control policies is to comparethe reductions in emissions achieved by various poli-cies with their costs. In its simplest form, this entailscomputing the cost of each control strategy and divid-ing the cost by the (weighted) reduction in emissionsachieved to yield a cost per ton of emissions reduced.The computation can indicate which options have thelowest cost per ton of pollution reduced and the sizeof the pollution reduction each option will achieve.Cost-effectiveness analysis can help by ranking

policy options in order of their incremental cost.

A more sophisticated approach computes cost perunit change in ambient pollutant concentrations, suchas cost per change of 10 micrograms per cubic meter(µg/m³) in ambient PM10 concentrations. This will in-volve modeling to estimate the impact of a reductionin emissions on ambient concentrations. This ap-proach enables comparison of different measures forreducing emissions from stationary, mobile, and fugi-tive sources.

The above can be taken a step further to computecost per life-year or disability-adjusted life-year(DALY) saved. A DALY is an indicator of the timelived with a disability and the time lost due to prema-ture mortality. In principle, this can be carried out notonly for air pollution reduction options but interven-tion measures in other sectors (such as supplyingclean water to households), and results ranked in or-der of decreasing cost-effectiveness. This approach re-quires giving relative weights to disability and mor-tality, but avoids having to assign monetary values tohuman health and life, which many find controver-sial. The problems encountered in going from changesin ambient concentrations to DALYs are discussed inannex 8.

Typically a number of distinct mitigation mea-sures can be pursued in parallel, and undertaking onewould not affect the cost or the effectiveness of otheroptions. For example, the Department of Buses of theMetropolitan Transportation Authority (MTA) New

York City Transit, which runs more than 4,000 buses,has been testing three different options to reducingexhaust emissions simultaneously: CNG, diesel-elec-tric hybrid, and “clean diesel” technology usingultralow3 sulfur diesel and particulate filters. In othersituations, for reasons of both cost and administra-tion, it may not be possible to evaluate multiple op-tions. For example, should new diesel trucks andbuses in São Paulo remain with Euro III emissionstandards or should they move quickly to meet morestringent Euro IV standards, which will require thatthe sulfur content of diesel be significantly lowered?In this case, cost-effectiveness analysis can be used tocompute the additional cost per ton of pollution re-duced of moving from a less stringent to a more strin-gent standard. Policymakers can then compare theadditional cost of moving from a less to a more strin-gent standard with perceived benefits to determinewhich policy to undertake.

Data requirements for cost-effectivenessanalysisPerforming a cost-effectiveness analysis requires thatthe analyst predict emissions for all sectors affectedby the policy with and without each policy.4 The diffi-cult step in a cost-effectiveness analysis comes in ag-gregating the emission reductions associated witheach policy, assuming that the policy reduces theemissions of more than one pollutant as is typicallythe case. Calculating the cost per weighted ton ofemissions reduced allows the analyst to rank pollu-tion control options by their average or by their incre-mental cost. Ideally, the weights assigned to reduc-tions in emissions of different pollutants shouldreflect the contribution of these emissions to somemeasure of ambient air quality (such as ambient PM2.5

concentrations) or to monetized benefits (as in a fullcost-benefit analysis). One of the greatest difficulties

3In this report, “ultralow” will be used to refer to sulfur levelslower than 50 parts per million by weight, corresponding to sulfurlimits to comply with Euro IV emission standards.

4In a cost-benefit analysis, emissions with and without thepolicy must be predicted for all sources of the key air pollutants inorder to produce the necessary inputs for air quality modeling.


lies in estimating weights to be assigned to differentpollutants, especially if ozone reduction is desired.

If cost-effectiveness is expressed in cost per DALY,then emissions have to be linked to ambient concen-trations and human exposure. The foregoing discus-sion on attributing sources to ambient concentrationsand their associated difficulties applies in this case.

It is also important to note that the use of cost-ef-fectiveness analysis to select instruments requires theprior specification of a constraint, either in the form ofan available budget to be expended to obtain thegreatest improvement in air quality or in the form of atarget level of ambient air quality to be achieved atleast cost. In either case, the determination of the con-straint will require political judgments about the costsand benefits of alternative uses of national resourcesto which the principles of cost-benefit analysis, dis-cussed earlier, will need to be applied.

Policy applicationsExamples of the application of cost-effectivenessanalysis to air quality policy assessment in develop-ing countries include those carried out in Jakarta(Grønskei and others 1996a), Kathmandu (Grønskeiand others 1996b), Manila (Larssen 1996a), andMumbai (Larssen 1996b) under the regional programURBAIR (Urban Air Quality Management Strategy inAsia). In each city, an emissions inventory was builtand rudimentary dispersion modeling carried out.Various mitigation measures for reducing PM10 wereexamined in terms of reductions in tons of PM10 emit-ted, the cost of implementation, the timeframe forimplementation, and health benefits and their associ-ated cost savings. Local stakeholders discussed andagreed on an action plan of abatement measuresbased on cost-benefit analysis, along with the identifi-cation of the lead agency, costs, time frame, and over-all priority. The abatement measures that have beenimplemented include introduction of unleaded gaso-line (and eventual elimination of lead in gasoline),tightening of standards, introduction of low-smokelubricants for two-stroke engine vehicles, vehicle-ex-haust emissions inspection to address gross polluters,and reduction of garbage burning. One of the associ-ated benefits was the creation of cross-sectoral work-

ing groups that met regularly. In Jakarta and Manila,these meetings are continuing under the Clean AirInitiative–Asia.

Marginal analysis and the “overlay” approachIn many circumstances transport policies do not haveair quality as their primary objective. But that doesnot mean that air quality should not be a consider-ation. In such cases an appropriate methodologywould be first to identify the policy or investment se-lected as optimal from the point of view of the trans-port benefits being sought. From this starting point,one could explore the cost, either in terms of in-creased financial outlay or lost transport benefit, of amodified policy or investment to yield a better airquality outcome than the base investment. Thatwould give an indicator of the cost-effectiveness ofthe adjustment to be compared with the costs atwhich those benefits could be achieved by an invest-ment or policy oriented toward direct air quality im-provement. This “overlay” approach has beenworked out in some detail for the energy and forestrysectors in the analogous case of GHG reduction strate-gies. (For more information, see Halsnæs and othersundated.) Countries would be well advised to insti-

tute a formal and systematic “air quality audit” of

transport policy and investment initiatives to iden-

tify cost-effective adaptations of such initiatives to

achieve collateral air quality benefits.

The Results of Analysis of AirPollution ControlAn important indicator of the value of air pollutionmeasures is the extent to which they can reduce mor-tality and morbidity impacts. The World Health Orga-nization estimates that urban air pollution and leadexposure together in 2000 accounted for 84 percent ofthe premature deaths from environmental causes inindustrial countries, but for fewer than 20 percent ofdeaths in developing countries as a whole and 10 per-cent for the low-income, high-mortality developingcountries where half of the deaths are due to unsafewater, sanitation, and hygiene. These proportions fallfurther if DALYs, which take morbidity into account


in addition to mortality, are used (WHO 2002). Trans-port is an important, but not the dominant, source ofurban air pollution in many developing countries.

Intersectoral comparisonsIt is informative to examine the health impact of out-door air pollution in the context of overall environ-mental health risks. WHO’s estimates indicate that,averaged across all developing countries, the impactof urban air pollution on premature mortality is about15 percent of the total number of premature deathsfrom all environmental health risks. This is less thanone-half of that from indoor air pollution and a littleover one-third of that from unsafe water, sanitation,and hygiene. The relative importance of the healthimpact of urban air pollution diminishes in both low-mortality and high-mortality developing countries ifDALYs are used as a metric rather than mortality. Fo-cusing only on environmental health risks as before,urban air pollution accounts for 6 percent of totalDALYs across all developing countries, 16 percent inlow-mortality and 3 percent in high-mortality coun-tries. But the importance of urban air pollution is in-creasing. In low-mortality developing countries, it isalready the second highest environmental health riskin terms of mortality and third highest in terms ofDALYs, and represents one-third of prematuredeaths. As countries take steps to increase access to

safe water and sanitation and cleaner household fuelsin the coming years, the relative importance of urbanair pollution is expected to increase.

When the same approach is applied to appraisalof options for tackling urban air pollution, the cost ofair pollution interventions expressed in terms ofDALYs can vary significantly as shown in table 2,with some of the possible actions outside the trans-port sector showing very high cost-effectiveness (lowcost per DALY or per life), particularly in the low-in-come, high-mortality countries. It would make sensefor these countries to select the measures with ex-tremely low cost per DALY first. As low-cost mea-sures are implemented and income rises, higher-costmeasures are adopted.

The short-term significance of diminishingreturnsPolicymakers will normally seek to give priority tomeasures that offer the greatest environmental benefitper dollar invested. For that reason, the earliest mea-sures adopted are likely to have the largest absoluteimpacts. For example, the U.K. government reportsthat compliance with Euro IV for gasoline vehicles in2005, which yields an extra 4 percent reduction inNOx and VOC, incurs an incremental cost of 200 Eu-ros (€200) per vehicle. This is nearly as much as thecost of removing the first 75 percent of NOx and VOC

Estimated Cost of Air Quality Mitigation Measures

Mitigation measure Cost in US$ Source

Improved stoves for reducing indoor air pollution, India 50–100 per DALY1 Smith 1998Switching from coal to gas in boilers in Katowice, Poland 2,900 per DALY World Bank calculationsSwitching from wood stoves to distillate fuel oil in Chile 3,100 per DALY World Bank 1994Diesel truck control (U.S. EPA 1991 standard) in Chile 5,997 per DALY World Bank 1994U.S. Tier 1 emission standards in Mexico 44,642 per DALY Lvovsky 2001U.S. Tier 2 emission standards in the USA 850,000 per life saved2 U.S. EPA 1999a

1Disability-adjusted life-year or DALY concerns the life years lost due to premature death and fractions of years of healthy life lost as a result of ill-ness or disability. One DALY can be thought of as one lost year of “healthy life.”

2The annual costs and number of lives saved are calculated for the year 2030. The costs (in 1997 US$) are $5.3 billion, and the number of livessaved 4,300. There are $1.84 billion in morbidity and other benefits that should be subtracted from the $5.3 billion. This implies a cost per lifesaved = $805,000. The median age of a person saved is 75, so that the cost per life-year saved is of the order of $100,000. One could argue thatthe benefits should not be subtracted at a value of $1.84 billion because they reflect U.S. morbidity values, which are high relative to those appli-cable in developing countries.

T A B L E2


by adopting Euro I in 1992. The cost of the latter cor-responds to the introduction of a three-way catalyticconverter. Meeting Euro II standards in 1996 cost anadditional €50 per vehicle corresponding to a further12 percent reduction of NOx and VOC. Meeting EuroIII standards in 2000 incurred an additional €400 andreduced NOx and VOC by another 6 percent. Thus thecost of removing that extra 6 percent from the exhaustwas substantially more than the cost of removing theprevious 87 percent (UK Commission for IntegratedTransport undated). A similar phenomenon of dimin-ishing returns is observable in respect to the impact ofprogressive reductions in sulfur content of fuels onparticulate emissions.

Another interesting comparison is to look at bothabsolute and percentage reductions in exhaust emis-sion levels. Urban buses in the United States were re-quired to emit no more than 0.80 grams of particulatematter per kilowatt-hour (g/kWh) through the 1990vehicle model year. This limit was lowered to 0.34 g/kWh for the 1991 model year, 0.13 g/kWh for 1993,0.094 for 1994, and 0.067 for the 1996 model year (seefigure 3). To achieve these reductions, the sulfur level

in diesel was lowered from 0.25 percent to 0.05 per-cent in 1993, with no further changes required since.In 2007, thanks to the introduction of ultralow sulfurdiesel and further advances in engine and exhaustaftertreatment technology, the limit will be reduced to0.013 g/kWh. In absolute terms, the most significantchange was from 0.80 to 0.34, a reduction of nearly 0.5g/kWh, from 1990 to 1991, relying only on enginetechnology improvement. The changes in the PMemission limit between 1990 and 1993 and between1993 and 2007 are broadly comparable in percentageterms, 83 percent in 1990–1993 and 90 percent in1993–2007. However, the 1990–1993 change gave anabsolute reduction in PM emissions of 0.67 g/kWh,more than five times the 1993–2007 reduction of 0.12g/kWh. The difference between the 1990 and 1996limits is even greater, 0.74 g/kWh. Moreover, thislarge reduction in PM emissions of 0.74 g/kWh be-tween 1990 and 1996 was achieved without loweringsulfur in diesel to ultralow levels: the maximum al-lowable sulfur limit remained at 0.05 percent between1993 and 1996. In contrast, ultralow sulfur diesel is anessential requirement for the 2007 limit, which gives a

U.S. Particulate Emission Standards for Urban BusesF I G U R E

3

Sources: U.S. EPA 1997a and 2000b.

Model year

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

g PM/kWh

0.25% sulfur

0.05% sulfur

0.0015% sulfur


reduction of 0.05 g/kWh over the 1996 standard. Lim-iting particulate emissions to a maximum of 0.013 g/kWh is a remarkable technological breakthrough andwill allow additional emission reductions in thosecountries and cities where ultralow sulfur diesel is in-troduced. However, for those many developing coun-try cities where the current emission levels are still at0.8 g/kWh or even higher, going down to 0.07 or even0.1 g/kWh—which can be achieved by lowering sul-fur in diesel to 0.05 percent together with engine im-provement and possibly an oxidation catalyst—canreduce transport-related air pollution considerablyand should not be discounted in the discussion ofcleaner fuels and emission reduction.

Given that many governments in low-income de-veloping countries are hard-pressed to commit sig-nificant investment resources to address urban airpollution issues, the immediate implication of dimin-ishing returns is that poorer developing countriesshould concentrate on finding ways of improving airquality that are most affordable and cost-efficient,given their current situation. That may mean focusingfirst on the measures already adopted in industrialcountries with large pollution reduction potential, asillustrated in the U.K. government report above.

Economic development and technological changeThe fact that there are short-term diminishing returnsin air pollution reduction is not the end of the story,however. As countries grow richer and their resourceconstraints become less pressing, the opportunitycosts of measures to improve air quality fall. Newtechnologies may be developed that reduce the abso-lute costs of securing specific reductions in emissions,so that as assets such as refineries or vehicles come torequire replacement, it becomes sensible to imple-ment higher standards. For all of these reasons lower-income developing countries should not accept re-taining old and less clean technologies, but shouldestablish the institutional framework for the progres-sive tightening of emission and fuel quality standardsin order to adopt emerging technologies as they be-come commercially available and affordable. Thequestion is thus not whether, but when and how newtechnologies can be introduced.

Appraising Instruments: A Structure forPolicy AppraisalThere are a number of aspects within which air pollu-tion reduction from mobile sources can be sought. Forpurposes of illustration, reducing urban air pollutioncan be disaggregated into three factors5 (figure 4):

� Less pollution per unit of fuel used� Less fuel used per passenger- or freight-ton-ki-

lometer traveled� Fewer (motorized) transport services required.

Policy instruments for dealing with these threefactors—reducing emissions per unit of fuel con-sumed, reducing the amount of fuel consumed per ki-lometer traveled, and reducing the number of kilome-ters traveled using motorized transport—are diverse.They can have effects on demand for or efficiency ofsupply of services, and can be implemented in a num-ber of ways, including legislation, infrastructure in-vestment, standard setting, and enforcement or fiscalincentives and disincentives. They involve govern-ment agencies in a wide range of sectors. Many policyinstruments address more than one component at atime. For example, optimal urban planning and landuse design may reduce the amount of motorizedtransport and, at the same time, decrease emissionsand increase fuel economy by helping minimize theamount of stop-and-start operations. Many of themeasures discussed are also most effective when theyare taken together; for example, fuels and vehiclesshould be treated as a joint system for proper opera-tion and emissions abatement.

The following three chapters discuss in turn eachof these factors for air pollution reduction, and therange of instruments that can be used to achieve

5The equation can, of course be expanded to emphasize par-ticular issues. For example, vehicle km traveled could be pre-sented as equal to the product of the number of vehicles and theaverage km per vehicle in order to highlight the importance oflimiting the fleet size. Or the emissions could be expressed perpassenger-km to give a better insight into the trade-offs betweenprivate motorized vehicles and well-patronized large public trans-port vehicles in containing emissions. The product of the first twoin many instances is better seen together, a good example ofwhich is assessing vehicle emission characteristics by comparingemissions per vehicle-km traveled. The three factors were selectedfor ease of discussing different dimensions of transport emissionsand are not intended to imply that fuels and vehicles should betreated separately or viewed in isolation from each other.


them. Where an instrument has effects on more thanone factor, it is usually discussed in the chapter whereits impact is seen as greatest, and cross-referencesmade. Mitigation measures affecting more than onecomponent simultaneously occur particularly withthe first two factors, so it is very difficult and often notpossible to judge which factor the measure would af-fect more. The two are shown as separate factors forconceptual simplicity, but should not be regarded assignificantly distinct in the type of policy responsesrequired or mutually exclusive. Some measures do

Factors Contributing to Transport Emissions

= ××Transport emissions

Emissions per unit of fuel

Total transportservices demanded

Fuel consumption per unit of transport service

F I G U R E4

not fall neatly into any of the three factors. One ex-ample is redirecting traffic away from environmen-tally sensitive areas. The objective in this case is to re-duce exposure, even if at the cost of increasing totaldistance traveled. Such measures with no obvious“home” are discussed under topics most closelylinked to them. Therefore, there can be considerablesubjectivity in assigning individual mitigation optionsto one of the three factors. Fiscal and institutional in-terventions are so all-pervading that they are treatedin chapters 6 and 7.

23

Reducing Emissions per Unit of FuelConsumed

Substantial improvements in air quality can be ob-tained through the introduction of cleaner fuels andvehicle technology, treated as a system. Regulationson vehicle emissions need to be set in harmony withregulations on fuel standards in light of the prevailingautomotive technologies, being set to allow satisfac-tory operation of vehicles while also limiting emis-sions (see annexes 1–4). Vehicles and fuels need to be

treated as a joint system for emission abatement.

Cleaner FuelsThe emission levels of two pollutants—lead andSO2—are directly related to fuel composition. Elimi-nating lead from gasoline, which is not naturallyfound in gasoline but is added to enhance octane, willeliminate lead emissions associated with fuel combus-tion from all gasoline-powered vehicles. For mostother pollutants, the emission levels tend to dependas much on the mechanical state of the vehicle andoperating conditions as on the fuel property.

Reformulating gasolineEliminating lead as an octane booster in gasoline is arelatively low-cost step with high returns in terms ofpublic health. Regulations to remove lead are typi-cally accompanied by controls on alternative ways ofboosting octane and by enforcing the use of catalyticconverters (to take advantage of the possibility of re-ducing CO, NOx, and hydrocarbon emissions dra-matically) for new vehicles. A great deal has beenlearned about how best to move to unleaded gasolinesince lead phase-out programs began in the 1970s. Inparticular, it is important that lead elimination be car-ried out in an integrated manner, considering the en-vironmental effects of alternatives to lead, the impact

on refinery upgrade schemes, and the effects on theexisting vehicle fleet. In the latter context, it is note-worthy that valve-seat recession, the only vehicle per-formance problem of concern potentially affecting oldvehicles switching from leaded to unleaded gasoline,has rarely been observed in practice, and certainly noton the scale anticipated based on engine laboratorytests. Elimination of lead from gasoline should be a

high priority for all countries that have not already

done so.

Most governments set standards for the composi-tion of transport fuels. In addition to the proscriptionof lead, controlled parameters for gasoline often in-clude benzene, sulfur, volatility, total aromatics, ole-fins, and oxygen. These measures reduce air pollutionin various ways (see annex 1 for more detail):

� Benzene is a carcinogen, and a fraction of ben-zene in gasoline is emitted unburned. Limitingbenzene in gasoline is a very effective way ofcontrolling benzene emissions, especially fromvehicles not equipped with catalytic converters.

� Sulfur reduces the efficiency of catalytic con-verters, by far the most effective means of re-ducing harmful pollutants emitted by gasolinepowered vehicles. Some of the state-of- the-artexhaust-control devices are extremely sulfur-in-tolerant, requiring the use of ultralow or “zero”sulfur fuels. As box 1 shows, the actual level ofsulfur in gasoline in developing countries can bevery low compared to regulated limits.

� Vapor pressure affects evaporative emissions.Reducing vapor pressure is one of the most cost-effective means of reducing the emissions oflight hydrocarbons (see the discussion on olefinsbelow).

C H A P T E R3


� Aromatics are a concern for two reasons. Somearomatics are photochemically reactive andstrong ozone precursors. Aromatics as a groupalso break down to form benzene in the engine,increasing benzene emissions.

� Olefins are strong ozone precursors. Conse-quently, in cities with high levels of ozone, con-trolling olefins in gasoline is likely to be impor-tant. Limiting gasoline vapor pressure is arelatively inexpensive way of controlling theevaporative emissions of light olefins.

� In vehicles with no oxygen sensors, the presenceof oxygen—in the form of alcohols or ethers—can facilitate combustion, reducing hydrocarbonand CO emissions if the vehicle is running“rich” (a low air-to-fuel ratio). This is an advan-tage primarily in old vehicles that do not havecatalytic converters and oxygen sensors. How-ever, some aldehydes formed from alcohol com-bustion are atmospherically active, potentiallycontributing to ozone formation. Oxygenateshave high octane, making manufacture of un-leaded gasoline easier. Oxygenates also “dilute”other undesirable components, such as benzene,sulfur, aromatics, and olefins.

Reformulating dieselThe characteristics of diesel that are of particular in-terest for air quality are density, sulfur, polyaromatichydrocarbons (PAHs), and cetane (see annex 1 formore detail). Fuel properties are often intercorrelatedand tend to move together. An example is density,aromatics, and cetane. It is therefore important todecouple fuel properties that change together beforeattributing the impact of each fuel parameter on emis-sions (Lee and others 1998). The impact of increasingcetane number on particulate emissions is engine-spe-cific. In the majority of cases cetane has little influ-ence, but in some cases particulate emissions increase,and in others decrease, with increasing cetane. Totalaromatics generally have no impact on particulateemissions. Decreasing density as well as polyaro-matics reduces particulate emissions in older technol-ogy, high-emitting vehicles, but has little or no impact

Actual Levels versus Limits

The actual level of a given fuel parameter may be quite different from

the legal limit, which cannot be exceeded at the pump. In some

cases, actual levels are substantially lower. For example, the gaso-

line sulfur limit in Pakistan is 1,000 parts per million by weight (wt

ppm) but the actual level is more than an order of magnitude smaller

because of the reliance on reformate (which contains essentially no

sulfur) as the main blending component. As such, gasoline in Paki-

stan may be classified as ultralow sulfur gasoline in practice. In the

United States, the current limit on diesel sulfur is 500 wt ppm, but

most fuels contain 200–300 wt ppm sulfur.

For fuels that have to meet a limit of 10 or 15 wt ppm at the

pump, the sulfur level at the refinery gate needs to be much lower—

possibly half of the legal limit at the pump—because of potential sul-

fur contamination during storage and transportation. If fuels with very

different sulfur levels are sold on the market, there is a serious risk of

cross-contamination.

B O X 1

on modern, lower-emitting engines. Back-end distilla-tion control (reducing the amount of fuel componentsthat boil at high temperatures) has little impact onparticulate emissions (Lee and others 1998). However,if the final distillation fraction consists of high-densityPAHs, reducing back-end volatility can be helpful, asfound with light-duty vehicles in the EuropeanProgramme on Emissions, Fuels, and Engine Tech-nologies (EPEFE) (ACEA and Europia 1995).

Sulfur occurs naturally in crude oil, and theamount of sulfur in “straight-run” diesel (diesel ob-tained from primary distillation of crude oil withoutfurther processing) is correlated with the crude sulfurcontent. The sulfur content of straight-run dieseltends to be much higher than that of unleaded gaso-line. During combustion, sulfur produces SO2, whichcontributes to secondary particulate emissionsthrough transformation in the air to sulfates. Sulfurreduction also has another advantage. The organic (asopposed to elemental) carbon component of particu-late emissions from diesel vehicles can be reduced bymeans of oxidation catalysts. Effective operation ofsuch catalysts requires sulfur in diesel to be no morethan 500 wt ppm and preferably lower, since catalysteffectiveness increases with decreasing sulfur. An-

CHAPTER 3. REDUCING EMISSIONS PER UNIT OF FUEL CONSUMED 25

other reason for limiting sulfur is that an oxidationcatalyst oxidizes sulfur and contributes to sulfate-based particulate emissions to varying degrees, de-pending on the catalyst activity (MECA 1999). The

level of sulfur in diesel is an important fuel parameter

affecting diesel emissions, and the target level should

be set in light of national circumstances and resource

availability.

Some reports have argued that lowering sulfurleads to lower engine corrosion and acidificationrates, and hence longer vehicle maintenance intervalsand lower maintenance costs, offering significant costsavings (GTZ 2002, Asian Development Bank 2003,Blumberg and others undated, ICCT undated). It isworth clarifying that these observations generally re-fer to lowering sulfur from thousands of ppm to 500ppm, rather than lowering sulfur from 500 wt ppm toan ultralow level (see annex 1, Diesel Quality Im-provement, for more detail). Acid formation is al-ready low at 500 wt ppm, and prolonged engine lifeon account of less corrosion is not expected whencomparing 500 wt ppm and 10–30 wt ppm sulfur. The

largest benefits of lower sulfur levels in terms of

lower engine corrosion and reduced maintenance

costs occur by reducing sulfur levels from thousands

of ppm to 500 ppm.

Several developing countries have already movedto a maximum of 500 wt ppm sulfur in diesel. Mexicoand Thailand have had a legal limit on automotivediesel sulfur of 500 wt ppm for a number of years. InJanuary 2004, the Philippines lowered the limit to 500wt ppm, while Thailand moved to 350 wt ppm. Oth-ers, such as Brazil, Chile, and India, have not yet lim-ited diesel sulfur to 500 wt ppm throughout the coun-try, but the limit on sulfur in large cities is 500 wt ppmor lower while the rest of the country is supplied withhigher-sulfur diesel, typically 1,500–2,500 wt ppm.The limit on sulfur in diesel in Santiago, Chile, is infact being lowered further from 300 wt ppm to 50 wtppm in 2004. Hong Kong, China, has switched en-tirely to 50 wt ppm sulfur diesel thanks to a signifi-cant tax differential between high- and low-sulfur die-sel, initially costing the government around US$0.10per liter (Ha 2002).

In contrast, many developing countries are still us-ing diesel fuel with a high sulfur content, in somecases as high as 7,000–10,000 wt ppm (or 0.7–1.0 per-cent). In these cases, the impact on particulate emis-sions of lowering sulfur to 500 wt ppm can be consid-erable, especially if particulate pollution ischaracterized by a high proportion of sulfates (see In-ternational Experience 2, next page). For those coun-tries with very high levels of sulfur in diesel where animmediate reduction to 500 wt ppm is difficult, aminimalist approach to secure consistency with Euro Ior U.S. EPA 1991 emission standards would requirelowering sulfur in diesel to 2,000 and 2,500 wt ppm,respectively. In some cases, this may be achievable atnot much additional cost (see annex 1 for when sulfurcan be reduced to these levels without much capitalinvestment). Even where levels of sulfur in diesel fuels

are very high, a strategy should be identified and

implemented to reduce sulfur to 500 ppm or lower as

a first step toward lower levels. In situations where

reduction to 500 ppm is very difficult in the near term

but lowering sulfur to 2,000–3,000 wt ppm is rela-

tively inexpensive, immediate steps should be taken

to lower sulfur to this level.

The cost of lowering sulfur in fuels is complex andrefinery-specific. In some cases, reducing sulfur to2,000–3,000 wt ppm is relatively inexpensive, but low-ering below that level is costly. In other cases, lower-ing to 10 wt ppm does not cost much more than low-ering to 500 wt ppm so that it would make sense tomove immediately to 10 wt ppm provided there is amarket for ultralow sulfur fuels. Examples of the im-pact of different refining situations in developingcountries on the cost of fuel sulfur reduction are givenin box 2 and annex 1. It should be borne in mind thathistorically, there has been a tendency among thosewho will be bearing the investment costs to use con-servative assumptions to estimate the cost of fuelquality improvement, resulting in relatively high fig-ures. These estimates tend to represent upper boundsand the actual costs may even be much lower. Agen-cies setting standards may make assumptions abouttechnology advances and associated cost reductionsin the future, giving lower figures than industry esti-


Diesel Sulfur Contribution to Emissions

The current debate on the merits and costs of mandating ultralow sulfur diesel in industrial countries has tendedto focus policymakers’ attention in developing countries on diesel sulfur to the exclusion of other considerations. Itis important to understand how sulfur in diesel affects emissions. Sulfur is emitted as sulfure dioxide and sulfurtrioxide, and transformed to sulfates in the atmosphere. Sulfates in turn can be combined with ammonia and othercompounds to form so-called secondary particulate matter. The extent of conversion to sulfates depends on anumber of factors, and the precise contribution of SO

x from mobile sources to secondary particulate formation

can be difficult to quantify, especially if there are other major sulfur emissions, for instance from coal combustion

or industrial processing.

In terms of tailpipe emissions, the early push to reduce sulfur in diesel arose from the need to lower the sulfatecontribution to particulate emissions. An illustrative example is given in the figure below, showing mass particu-late emissions from new heavy-duty diesel vehicles in the United States as a function of vehicle technology anddiesel sulfur. In 1988, the majority of particulate emis-sions were carbon based, so that reducing sulfur wouldnot have enabled large overall reductions in particulateemissions. By steadily improving vehicle technology,however, vehicle manufacturers achieved significant re-ductions in carbon-based particulate emissions. By the1994 model year, sulfates constituted more than half ofparticulate mass, at which point it was not possible tomeet the new diesel emission standards without reduc-ing diesel sulfur further. Therefore, in 1993, the U.S.EPA mandated sulfur in diesel to be lowered by a factorof five from 0.25 percent by weight (2,500 wt ppm) to0.05 percent (500 wt ppm). The European Union fol-lowed in 1996.

In countries where diesel sulfur is in the thousands of ppm and the particulate composition looks similar to the1988 profile shown in the above figure, the first priority would be the reduction of carbon-based particles. For ex-ample, overfueling, a common practice in many developing country cities, increases the so-called black carbonemissions markedly, so that exploring measures to stop this practice can reduce particulate emissions (Kweonand others 2002). As steps are taken to reduce the carbonaceous components, diesel sulfur reduction can be pur-sued in parallel, targeting 500 wt ppm, if not lower. Some maintenance savings can be achieved by lowering sulfurfrom high levels, as the presence of sulfur in diesel leads to acid formation, corrosion of vehicle parts, and acidifi-cation of lubricants, requiring more frequent oil changes and shortening the engine life. Going down from thou-sands of ppm to 500 ppm will also help reduce ambient secondary sulfate particulate concentrations significantly.

Below 500 wt ppm, the direct benefit of further sulfur reduction on particulate emissions is small (Lee and others1998, Asian Development Bank 2003). Because acid formation is already low at 500 wt ppm, vehicle maintenancebenefits are also expected to be small (Weaver 2003, Lowell 2003). Lowering sulfur markedly below 350–500 wt ppmtherefore makes sense primarily if the use of extremely sulfur-intolerant advanced exhaust control devices is planned.*

*The first program in developing country cities to retrofit diesel vehicles with particulate filters was in Hong Kong, China. The retrofitting of me-chanical particulate filters in Hong Kong did not require diesel sulfur reduction (although the government provided a subsidy of US$0.10 per liter toswitch to 50 ppm sulfur diesel) because there was no catalytic component in the trap. In the absence of automatic regeneration, these filters requirewashing as frequently as every two to three days (Ha 2002, Tsang and Ha 2002).



Source: McCarthy 1994.

Particulate Emissions from New U.S. Heavy-Duty Diesels

0

0.1

0.2

0.3

0.4

0.5

0.6

1988 1991 1994 1994

g PM/horsepower-hour

Carbon soot & organicsSulfate

Model year

0.25% sulfur

0.05% sulfur


mates. In addition, it is not uncommon for differentagencies in the government—for example, environ-ment and energy—to come to significantly differentnumbers for cost estimates.

The EU and North America are planning to movetoward 10 (also called “sulfur-free”) and 15 wt ppmsulfur limits in diesel, respectively, during the latterhalf of this decade, in order to take advantage of theadvanced diesel exhaust emission reduction technolo-gies that have extremely low tolerance for sulfur.These include particulate filters (traps), NOx

Cost of Fuel Reformulation: Examples from Latin America and the Caribbean and from AsiaB O X 2

The Latin American Energy Organization (OLADE) and the Regional

Association of Oil and Natural Gas Companies in Latin America and

the Caribbean (ARPEL) undertook an extensive study over two years to

examine the refining sector in the region, anticipated demand and fuel

quality changes, capital requirements, and related financial needs

(ESMAP 2002b). For the purpose of the study, the Latin America and

the Caribbean region was divided into four subregions. Two refined pe-

troleum product demand growth scenarios were considered (high and

low), and the product quality specifications included 2.5 percent ben-

zene and 400 wt ppm sulfur in gasoline and 500 wt ppm sulfur in diesel

in the reference target case. In addition, the option of reducing sulfur to

50 wt ppm was studied separately for region 3 comprising southern

Brazil, Argentina, Uruguay, Paraguay, Chile, and eastern Bolivia.

In the high-demand scenario, going to the reference target case

from the current situation entailed an investment of US$6.5 billion to

meet the fuel quality specifications, and an additional $27.7 billion to

meet product demand growth. The corresponding estimates for the low-

growth scenario were $3.6 billion and $8.4 billion, respectively. In the

step-out case for region 3, environmentally driven fuel quality changes

including lowering sulfur to 500 wt ppm incurred an additional capital

cost of $2 billion, but going down to 50 wt ppm increased the cost by an

additional $4 billion to $6 billion (both costs reflect making fuel quality

changes in one step and are for the high-growth scenario).

As with other similar refinery studies, simplifying assumptions

made in the study have a large impact on the final cost estimates. As

the study reports explain, construction costs were set equal to those in

the U.S. Gulf Coast. In practice, experience in the region indicates that

costs can be 40 percent higher. The use of an aggregate refinery model

(one notional refinery was set up for each region) results in efficiencies

that may not be achievable by individual refineries, which could under-

estimate capital costs by 10–20 percent. The need to process increas-

ingly heavier crude oils in the future could also add to investment costs.

Future advances in refinery process technologies, however, are likely to

lower investment and operating costs significantly. Another factor that

could lower the overall investment costs is optimization of intra-regional

trade.

The potential costs of sulfur reduction have been studied else-

where. As might be expected, the studies vary in purpose, focus, and

level of detail. Some apparently low costs (0.4 U.S. cents per liter for

gasoline and 1 cent per liter for diesel) have been quoted as being rep-

resentative of cost figures in developing countries, based on a study of

the refining sector in China (Yamaguchi and others 2002, ICCT un-

dated). Based on these numbers, the cost of going in one step to 10

ppm gasoline and diesel has been claimed to be affordable even in de-

veloping countries (Blumberg and others undated). But the study on

China was undertaken to indicate refinery process options available to

China to produce low-sulfur fuels. The cost figures are based on refin-

ery equipment costs only, and do not include operating and off-site

costs, which could add 25–50 percent more to the costs according to

the study report. Moreover, the capital cost figures are for standard-

sized units at large refineries, suggested by the commissioning agency.

Many of the capital cost assumptions were a fraction of capital costs

experienced in the rest of the world. Verifying the capital cost figures

was outside the scope of the study. In reality, there are many relatively

small refineries in China for which investing in advanced sulfur removal

technology would not be feasible. The low costs cited tell only part of

the story, and therefore the study’s authors made no claim that the

costs were a fully built-up per-liter cost of producing sulfur-free fuels

even in China, much less for other developing countries. Each market

will have unique characteristics in terms of demand levels and patterns,

crude resources, technology in place, prices, and capital and operating

costs (Yamaguchi 2003).

adsorbers, and selective catalytic reduction (see an-nexes 2 and 3).1 These devices can make diesel ve-hicles essentially as clean as CNG vehicles. The com-bined use of ultralow sulfur diesel with particulatefilters (currently suitable primarily for heavy-dutydiesel vehicles, with advances being made to makefilters robust for light-duty) and other advanced ex-

1Three-way catalysts capable of converting NOx do not oper-ate at high air-to-fuel ratio (also called “lean”) typical of diesel en-gine operation. A lean de-NOx catalyst is capable of convertingNOx even when the air and fuel mixture is lean.


haust control devices is collectively called “clean die-sel,” in contrast to conventional diesel. A corollary ofrequiring diesel sulfur to be lowered to ultralow lev-els should be to simultaneously require new diesel ve-hicles to meet extremely stringent particulate emis-sion standards that require the use of particulatefilters for PM control. If not, the benefits reaped fromthe large investment required to lower sulfur in dieselcould be disproportionately small.

Reducing sulfur to 10–15 wt ppm in turn requiresexpensive hydrodesulfurization processes, althoughinvestment costs are expected to fall with increasingscale of application and experience. The incrementalcost of producing ultralow sulfur fuels is generallythe lowest in large-scale refineries. Where large refin-

eries are being built or upgraded, countries should se-

riously consider investing in the necessary facilities

to produce ultralow sulfur fuels. Importing countries

can take advantage of ultralow sulfur fuels as they

become increasingly available. For other non-import-

ing developing countries, the cost-effectiveness of

mandating ultralow sulfur fuels should be compared

with other strategies for improving air quality.

Harmonizing standardsGiven the existence of substantial scale economies inmanufacturing, harmonization of standards may re-duce production costs of vehicles and fuels andstimulate regional trade. Vehicle manufacturers aretherefore leading initiatives to harmonize fuel qualitystandards worldwide in accordance with vehicleneeds. More specifically, European, North American,and Japanese automobile and engine manufacturersassociations with their associate members around theworld have issued the World-Wide Fuel Charter(ACEA and others 2002), which calls for consolidationof fuel specifications into four categories (see annex4). Category 1 is the least stringent while Category 4 isdesigned for the most advanced vehicle technologies.Trade considerations (between Canada and theUnited States, within Central America, and within theEU) are a strong driving force for harmonizing fuel-and vehicle-emission standards. Fuel specifications inthe EU and North America are already integrated,and similar measures have been proposed in Latin

America and the Caribbean. In the former SovietUnion republics, fuel standards are already largelyharmonized by virtue of the countries’ history.

Countries that import the bulk of their transportfuels will find it easier to harmonize with those neigh-boring countries from which they import than willcountries that have domestic refineries. However, iftwo neighboring countries (for example, India and SriLanka) have very different air pollution problems, itmay not make sense to harmonize fuel- and vehicle-emission standards, unless the country with less seri-ous air pollution problems is already importing themajority of fuels and vehicles from the other. Devel-oping countries may benefit from regional harmoni-zation agreements. At the same time, it is necessary totake into account the key transport pollution prob-lems in individual countries, and the infrastructurecapabilities of refining and distribution in the region.

One of the objectives of standard harmonization isto minimize the number of fuel types and vehicleemission requirements. From the point of view ofmanufacturers, the fewer the number of standards,the greater the potential for trade and for taking ad-vantage of scale economies. That said, especiallywhen harmonizing standards across a large region, itis inevitable that some countries have the resourcesand air-quality-based needs to impose tighter stan-dards earlier than others. Early introduction ofcleaner fuels by some member states is therefore com-mon in harmonizing standards.

Regionally differentiated standardsAnother approach is to introduce regionally differen-tiated or city-specific standards whereby more strin-gent standards are imposed on large metropolitan ar-eas. This approach has been used in Brazil, Chile,India, and Mexico, and earlier in the former SovietUnion where leaded gasoline was banned in citieswith a population of more than 1 million beginning inthe 1980s. If tightening standards throughout thecountry is too costly in the near term, provided thatthe distribution system can handle segregation of fu-els, it could be more cost-effective to supply cleanertransport fuels to large cities only, and confine the useof fuels with less stringent specifications to areas out-


How do you decide when to lower transport fuel sulfur limits, and to what level?

The issue that dominates the debate on fuel quality in North America

and Europe is limits on sulfur in gasoline and diesel. As a result, diesel

fuel quality improvement has come to be synonymous with sulfur re-

duction in the minds of many policymakers.

In considering the levels to which sulfur in fuels should be limited, it is

important to ask how sulfur reduction affects emissions, concentrating

on diesel, because sulfur in diesel is a more serious problem in devel-

oping countries than is gasoline. In developing country cities, the most

important pollutant to target is generally particulate matter. For a given

diesel vehicle technology that does not rely on an oxidation catalyst,

the main impact of reducing sulfur in diesel is to reduce the sulfate con-

tribution of particulate emissions, including secondary particulate for-

mation. For developing countries with high sulfur diesel, it would make

sense, even in countries with refineries, to lower sulfur in diesel to at

least 2,000–3,000 wt ppm (see annex 1 for more detail) as an immedi-

ate step and preferably to 500 wt ppm or lower. A sulfur limit of 500 wt

ppm is needed to enable U.S. 1994 and Euro II emission standards. At

sulfur levels in the upper tens of thousands of ppm, the sulfate share of

total particulate emissions is likely to be significant, and there are indi-

cations that sulfur-containing particles are especially harmful. But

where the carbonaceous contribution to particulate emissions domi-

nates, as is often the case in the majority of heavily polluting diesel ve-

hicles in developing countries, reducing sulfur alone may have a limited

impact. Lowering density may be just as important in those circum-

stances.

The impact of sulfur reduction on particulate emissions from diesel ve-

hicles is significant down to 500 wt ppm, below which the direct ben-

efit (that is, without particulate filters) in terms of particulate emissions

reduction becomes minimal (Lee and others 1998). For countries with

refining capacity, reducing sulfur to 500 wt ppm almost always requires

refinery upgrade, and the capital cost depends on such factors as the

size of the refinery, availability of hydrogen from other processes, and

the type of crude processed. Units for hydrotreating to reduce sulfur

can cost tens of millions of dollars and require an adequate source of

hydrogen. For countries that import some or all of their diesel fuel, the

decision is simpler: the primary question is whether the incremental

cost of purchasing lower sulfur diesel can be justified in terms of ben-

efits and passed on to consumers.

The next important threshold is 50 wt ppm, below which particulate fil-

ters can be considered. This is the decision facing those developing

countries and cities that have already moved to 500 wt ppm sulfur. The

primary objective of ultralow (below 50 wt ppm, and preferably 10 or 15

wt ppm) sulfur limits is to meet extremely stringent particulate and NOX

standards simultaneously (for trade-offs between particulate and NOx

reductions, see the section in this chapter, Vehicle Technology). Man-

dating ultralow sulfur diesel makes sense only if (catalytic) particulate

filters are also mandated for new vehicles. For those developing country

cities that are willing and able to pay for particulate filters and dramatic

sulfur reduction, this step is well worth considering.

There are teething problems in commercializing new technology. For

example, while applications of diesel particulate filters have been

mostly successful in the United States, a few problems have been en-

countered. In one case, engines equipped with exhaust gas recircula-

tion (EGR) and particulate filters sold by a leading U.S. diesel engine

manufacturer experienced problems with filter clogging, even though

the technology supplier was actively involved in providing technical ad-

vice. This is not unlike the problems encountered with early-generation

CNG buses. Waiting until many of these problems have been worked

out may be a sensible option for those developing country cities with

limited experience with maintaining modern technology vehicles. The

technology to achieve reductions to 10–15 wt ppm, as well as the ex-

haust-control technology enabled by ultralow sulfur diesel, is undergo-

ing continuous improvement. As a result, significant cost reductions are

almost certain in the coming decade.

F A Q 1

side urban centers with less serious air pollutionproblems. As increasingly stringent fuel quality andvehicle emission standards are contemplated, this tar-

geting of large metropolitan areas for tighter stan-

dards, so that refineries have a longer period in which

to upgrade their process units and have an outlet for

lower-quality fuels during the interim, may present a

cost-effective alternative to tightening standards

throughout the country all at once.

The logistics of delivering fuels of different quali-ties to different depots, as well as leakage and otherenforcement problems, have made it difficult toimplement regional differentiation cost-effectively inthe past. For long-distance transport vehicles, refuel-ing with different quality fuels in different parts of thecountry may preclude the use of fuel-sensitive ex-haust control devices. These concerns notwithstand-ing, requiring that vehicles in large polluted cities pro-


viding intra-city passenger and goods transport meetmuch tighter emission standards and the entire city issupplied only with cleaner fuels to enable them to doso merits serious consideration.

Downstream petroleum sector reformOne of the important factors for making sustainableimprovements in fuel quality is an efficient downstreampetroleum sector. Where there are serious distortionsin the sector, unsustainable subsidies, inefficiencies, ashortage of investment, or a lack of maintenance ofexisting assets, coupled with the protection of the sec-tor, it is very difficult to realize significant fuel qualityimprovements. Fuel quality is often directly related

to the efficiency of the downstream petroleum sector.

This is especially problematic in countries with oilrefineries. In a number of developing countries, thegovernment is the sole owner of domestic refineries,which it protects by means of import restrictions, quo-tas, or high tariffs. In many cases, mounting govern-ment subsidies and the inability to attract capital forneeded investment in the sector are two primary driv-ers for sector reform. Environmental considerationshave typically played a minor role. While environ-mental degradation may not drive sector reform, fuelquality improvement cannot often be divorced fromit, and in a number of circumstances sector reform is

in fact a prerequisite for fuel quality improvement.

In countries with inefficiently operated domesticrefineries, a further consequence of protection of theserefineries is that consumers are denied access tocleaner fuels that are available internationally and areoften cheaper than dirtier, domestically produced fu-els. As fuel quality specifications improve worldwide,it will become increasingly difficult for inefficient orsmall refineries to produce fuels that meet the stan-dards without requiring even greater protectionagainst competitively priced and cleaner fuels fromoutside the market. For example, the number of refin-eries in North America and Europe has declined inthe last two decades as the downstream petroleumsector has become liberalized, and small or inefficientrefineries have closed down in the absence of govern-ment protection. Fuel quality specifications in coun-

tries with protected, state-owned refineries tend to

lag behind those of countries where the petroleum

sector is more open.

Even where the government does not have sharesin the ownership of refining capacity, it can still beheavily involved in the sector, setting prices at variouspoints in the supply chain, deciding who can importwhat and how much, and determining how fuelsshould be transported throughout the country. Thatdenies the country the benefit of market signals thatcan help allocate resources efficiently. Even if there areno refineries within the country, if prices are con-trolled by the government or, worse, significant pricesubsidies are given to make fuels cheaply available toconsumers, improving fuel quality, which entails costincreases, becomes problematic for the country. Shortof raising fuel prices, government subsidies will in-crease. Raising fuel prices becomes highly politicizedunder these circumstances, making it difficult to passthe incremental cost to consumers. The alternative—carrying an even larger subsidy bill—is typicallyequally unacceptable to the government. Large subsi-dies also invite illegal diversion, thereby greatly re-ducing the access of the intended beneficiaries to thesubsidized fuels. At a minimum, effective benefitsand possible adverse environmental effects of fuel orrefining subsidies need to be carefully considered. Acompetitive market reduces opportunities for corrup-

tion and provides a sound basis for attracting new in-

vestment without creating contingent liabilities for

government or requiring the pledging of public assets.

Establishing fair, healthy, and transparent compe-tition in the supply of fuels to the domestic marketand market-based fuel pricing is thus an integral ele-ment of the policy for introducing cleaner fuels. Aneffective and well-regulated competitive market im-poses relentless pressure on participants to improveefficiency and to share the gains with customers.Given vigorous competition on the world market, im-

porting superior quality petroleum products can be

the most cost-effective option for many developing

countries.

While refining, importing, and distributing areoligopolistic in nature because of the large invest-


ments involved, international experience has shownthat it is possible to achieve effective competitionwhen individual players have 20 to 30 percent of thetotal market. For countries where the downstream pe-troleum sector is not yet open, restructuring the sectorto introduce effective competition is a first step. Tothis end, governments should adopt licensing andregulatory regimes that provide a level playing field.Competitive efficiency will usually be enhanced by

opening product imports, storage, distribution, and

retailing businesses to new entrants.

Within a liberalized fuel sector it becomes increas-ingly important for government to have a clear strat-egy for taxes and fiscal incentives. Many countrieshave used tax discrimination to assist the eliminationof lead from gasoline, and some, such as some EUcountries, are now using such incentives to acceleratethe reduction of sulfur in diesel. Among developingcountries, Hong Kong, China, introduced in July 2000a net subsidy of US$0.10 per liter in favor of dieselcontaining 50 wt ppm sulfur (Ha 2002). Because this

subsidy made the retail price of 50 wt ppm sulfur die-sel lower than that of regular diesel (at the time up to500 wt ppm sulfur was allowed in regular diesel), theentire diesel retail sector switched to 50 wt ppm.However, a subsidy of US$0.10 per liter is very highby any standard. The high cost of this subsidy was inturn largely due to the extremely limited availabilityof 50 wt ppm sulfur diesel in the region at the time.When significant implicit or explicit fuel subsidies

are being considered, governments should compare

the likely benefits with alternative uses of the funds,

including the option of waiting until international

prices come down.

For all countries, the prices and quality of refinedproducts sold in neighboring countries are importantconsiderations in trying to enforce high domesticstandards. If cheaper and inferior quality fuels arereadily available along the borders, it becomes moredifficult to tighten fuel specifications and enforcethem. Similarly, widely different taxes on different fu-els that are to some extent substitutable (for example,

If it costs only a cent or two a liter to improve fuel quality, why can’t we improve fuel qualityimmediately?

F A Q 2

When translated to U.S. cents per liter, fuel quality improvement costs

seem quite modest. This is especially true in the case of gasoline in

countries where gasoline is already heavily taxed. Motorists, especially

private car owners who tend to be from upper income groups in devel-

oping countries, may not even notice whether gasoline costs 70 or 71.5

cents per liter. Certainly, if a country relies on imports to meet part or all

of its fuel demand, it makes sense to import higher-quality fuels for use

in large cities, if not in the entire country. Several developing country

governments have successfully managed regional differentiation of fuel

quality, such as the early introduction of Euro II-compliant fuels in

Chennai, Delhi, Kolkata, and Mumbai in 2000–2001 in India, some 4–5

years ahead of the rest of the country, and tighter fuel quality standards

in Mexico City compared to the rest of Mexico.

For developing countries with domestic refineries, the first, and perhaps

the most serious, barrier to fuel quality improvement is the need to

raise financing for refinery revamp projects. One estimate of the cost of

installing a diesel hydrotreater to reduce sulfur to below 500 wt ppm at

a 50,000 barrels per day refinery in South Asia is $73 million (1998

US$) (ESMAP 2001b). Improving fuel quality without changing the

product output does little to increase refinery profits. The return on in-

vestment to upgrade fuel quality is often very low. Under these circum-

stances, it is difficult to raise financing (foreign or domestic) for even a

well-run refinery. The situation is much more difficult for an inefficiently

run or small-scale refinery in a country with government price controls.

This inability to raise financing has led to continuing use of leaded

gasoline in some countries and high-sulfur diesel in a number of others.

As far as diesel is concerned, many developing countries levy a low tax

rate to keep the end-user diesel price low. This makes the incremental

cost of improving fuel quality larger as a percentage of the retail price

compared with that of gasoline. Because the majority of road diesel is

used in goods and public transport, an increase in the price of diesel

has economy-wide repercussions. This in turn makes policymakers

hesitant to raise the tax on diesel or adopt measures that will raise the

price of diesel. These impacts can be lessened considerably without

much reduction in environmental health benefits if cleaner fuels are tar-

geted only in large metropolitan areas.


gasoline and naphtha for petrochemical production)make it difficult to enforce standards.

Maintaining Fuel StandardsFuel quality standards are ultimately driven by themaximum vehicle emission levels to be permitted.Setting standards makes sense only if the governmentintends to enforce them, and enforcement can be diffi-cult and costly. Take the case of emission standardsfor new gasoline vehicles. After lead elimination,many governments have rightly required that newgasoline vehicles be fitted with catalytic converters.But if adulteration of gasoline with (much lower-priced) kerosene occurs as a matter of routine, prema-ture catalyst deactivation can occur. The cost of com-

pliance and the probability of compliance are key

questions in setting standards.

Adulteration of fuelsAdulteration of standard fuels (World Bank 2002d)consists of the addition of other components to thebase fuel. Adulteration typically occurs when eitherthe seller or the purchaser of fuel sees some advan-tage to be taken, either in terms of lower costs of ve-hicle operation or greater profit from adulteration.This advantage arises from two main sources:

� Adulterants that have little market value (for ex-ample, waste industrial solvents or off-specifica-tion petrochemicals)

� Additives that reduce the cost of the fuel be-cause of tax differences (for example, untaxedfuels for exports or subsidized kerosene).

Even if fuels meet specifications at the port of en-try or the refinery gate, it is all too common for themto be adulterated somewhere in the supply chain(Kojima and Bacon 2001a). Examples include illegaladdition of cheap lead additives from the RussianFederation to gasoline in Central Asia, addition ofmuch lower-priced kerosene to gasoline in manycountries, and addition of off-specification chemicalsto gasoline and of heavier fuels to diesel. In a largenumber, if not the majority, of developing countries,gasoline carries a much higher tax than diesel. Kero-

sene is often untaxed, or even subsidized, as a basiccooking and lighting fuel for the poor, and hence canbe used to reduce the cost of auto fuel. Industrial sol-vents and recycled lubricants are other materials withlittle or no tax, inviting adulteration. Financial gains

arising from differential taxes or price controls are

the primary cause of fuel adulteration.

For gasoline, any adulterant that changes its vola-tility can affect drivability. High volatility (resultingfrom the addition of light hydrocarbons) in hotweather can cause vapor lock and stalling. Low vola-tility in cold weather can cause starting problems andpoor warm-up. The most common form of adultera-tion that is damaging for emissions is adding kero-sene to gasoline. Kerosene is more difficult to burnthan gasoline, so that its addition results in higher lev-els of hydrocarbon, CO, and particulate emissionseven from catalyst-equipped cars. The higher sulfurlevel of kerosene can also reduce the efficiency of thecatalyst, thereby lowering catalytic conversion of en-gine-out pollutants. If too much kerosene is added,volatility is lowered and, more seriously, octane qual-ity can fall below the octane requirement of the en-gines and engine knocking can occur. Besides possiblydamaging the engine mechanically, knock can in-crease particulate, hydrocarbon (HC), and NOx emis-sions. The addition of kerosene to gasoline increases

emissions under all circumstances and should be dis-

couraged by close monitoring and other means.

For diesel, the most common form of adulterationis also kerosene addition. This is not always harmful.The blending of kerosene into automotive diesel fuelis in fact widely and legitimately practiced by the oilindustry worldwide as a means of adjusting the low-temperature operability of the fuel. This practice isnot detrimental to tailpipe emissions, provided the re-sulting fuel continues to meet engine manufacturers’specifications (especially for viscosity, cetane number,and sulfur); in some respects it may even reduceemissions. However, high-level adulteration of lowsulfur (for example, 500 wt ppm) diesel fuel withhigher-level sulfur kerosene may cause the fuel to ex-ceed the sulfur maximum. Further, the keroseneblended into diesel may have poorer lubricity and


promote fuel system wear, higher flammability, andpotentially a lower cetane number, and producehigher NOx emissions. The addition of heavier fueloils to diesel is usually easy to detect because the fuelwill be darker than normal. Depending on the natureof these heavier fuel oils and the possible presence ofadditional PAHs, there could be an increase in bothexhaust particulate and PAH emissions.

Where extremely sulfur-intolerant exhaust controldevices are in use, the adulteration of gasoline or die-sel with higher-sulfur fuels could be very damaging.If kerosene containing 1,000 wt ppm sulfur is addedto diesel containing 10 wt ppm sulfur, the blend couldpermanently deactivate sulfur-sensitive devices. Evenwith less sulfur-sensitive devices, the addition of, forexample, railroad diesel containing 5,000 wt ppm sul-fur to automotive diesel containing 350 wt ppm sulfurcould result in considerable harm.

It is possible to adulterate fuels and still complywith fuel standards. When gasoline is judiciouslyadulterated with gasoline boiling range solvents suchas toluene, xylenes and other aromatics, or light mate-rials such as pentanes and hexanes (rubber sol-vents)—available at low or zero tax for their normalindustrial use—the gasoline may continue to meet allspecifications and not exhibit drivability problems.But larger amounts of toluene or mixed xylenes couldcause an increase in hydrocarbon, CO, and NOx emis-sions and significantly increase the level of air toxins,especially benzene, in the tailpipe exhaust. Extremelyhigh levels of toluene (45 percent or higher) couldcause premature failure of neoprene, styrene butadi-ene rubber, and butyl rubber components in the fuelsystems. This has caused vehicle fires in some parts ofthe world, especially in older vehicles. The adulter-ated gasoline itself could also have increased poten-tial human toxicity. Adulteration of gasoline by wasteindustrial solvents is especially problematic becausethe adulterants are so varied in composition. Adulter-ants may contain halogens, silicon, phosphorus, orother metallic elements (found in recycled lubricants);these are quite outside the normal gasoline composi-tion range. They will cause increased emissions andmay even cause vehicle breakdown by corroding fuelinjection systems and carburetors and by causing de-

posits on valves, fuel injectors, spark plugs, oxygensensors, and exhaust catalysts. Even low levels ofadulterants can be very injurious and costly to the ve-hicle operator. Adulteration of gasoline with cheaper

fuels and waste industrial solvents can be highly

dangerous.

In cases where adulteration does not increaseemissions, it may still be socially undesirable becauseof its indirect effects. For example, large-scale diver-sion of rationed kerosene subsidized for householduse to the diesel automotive sector does not necessar-ily increase emissions from diesel vehicles, but de-prives the poor of kerosene for cooking. Lack of avail-ability of subsidized kerosene forces the poor tocontinue to use biomass and may expose them to highlevels of indoor air pollution. Protection of the supply

of subsidized cooking and lighting fuel to poor house-

holds calls for more vigorous enforcement of the ban

on the use of rationed kerosene as an automotive fuel

adulterant.

When engines are out of tune with the specifica-tions set by the manufacturers or are poorly main-tained, they will emit substantially more pollutants—even when operating on fuels that meet allspecification requirements—than properly main-tained vehicles. Whenever considering the impact offuel adulteration on air quality, it is important to keepthe impact of adulteration in perspective. The effects

on emissions of basic engine design and maintenance

usually far outweigh changes in fuel composition.

Enforcing fuel standardsIdentifying and checking fuel adulteration presents adifficult challenge in the face of enormous incentivesto adulterate in some countries. As vehicle emissionstandards are progressively tightened and fuel qualityand vehicle technology are increasingly integrated,having fuels that meet the specifications becomes allthe more important for meeting the new emissionstandards. It is one thing to prohibit adulteration offuel in developing countries and quite another matterto enforce it. Governments need to have well-orga-nized procedures to detect and eliminate adulteration.

An important step in tackling fuel adulteration isto reduce the incentives to engage in it. These depend


on the relative benefit (from adding low-priced mate-rials) and cost (from the risk of being caught andfined, losing one’s market share on account of declin-ing reputation, or having one’s business license re-voked). The benefit arises from differential taxation,tax evasion, and lower production costs of adulter-ants. Though tax incentives can always be reduced,fiscal policy has multiple objectives: concerns aboutfuel adulteration, however serious, cannot be the soledriver of fiscal policy. The cost to those engaged inadulteration depends on the ability of the regulatingauthorities to detect adulteration and to impose suffi-ciently punishing sanctions to deter recurrence of fueladulteration. Government policies on fuel quality en-

forcement are likely to be most effective if they com-

bine the reduction of incentives to adulterate with

measures to detect and punish criminal adulteration.

A number of analytical techniques are available todetect adulteration: density measurements, determi-nation of evaporation and distillation properties, hy-drocarbon composition analysis by gas chromatogra-phy, and use of trace markers in fuels. Fieldtechniques produce results quickly and conveniently,but are not as detailed or quantitatively accurate aslaboratory tests. For the majority of the tests, accuratedata on uncontaminated fuels are also a prerequisite.In all cases, it is important to have good sampling

techniques and access to a good petroleum analytical

laboratory.

Fuel adulteration has significant financial benefitsfor those engaged in it. Therefore, anyone investigat-ing it must have the direct protection of local law en-forcement agencies in taking field samples. Even inthe best of circumstances, taking and maintainingsamples for checking fuel quality is not easy. In devel-oping countries, finding proper sample containersand avoiding harassment at retail outlets while sam-pling are real operational problems. These problemsare compounded by lack of experience in checkingspecifically for adulteration. Precision and repeatabil-ity could be improved by setting up programs forcross-checking inter-laboratory variability. This re-quires action by the responsible agencies within gov-ernment. The first step is recognition of the existenceand seriousness of the problem—both from the con-

sumer and environmental perspectives—by suffi-ciently high levels of government. Given historicalproblems with fuel adulteration in many developingcountries, strong political will needs to be generated

before appropriate regulatory and enforcement steps

can be taken.

The manner in which retail fuels are distributedhas an important bearing on fuel adulteration. For ex-ample, having large numbers of small, independenttransport truck operators moving fuels from termi-nals to the point of sale creates an environment con-ducive to adulteration. Adulteration may also occurwith the collusion of the retail outlet operator. If gov-ernment officials are involved in adulteration, estab-lishing a good monitoring and enforcement mecha-nism becomes all the more difficult. An effective

market-based approach is practiced in many indus-

trial countries where oil companies market at retail

and assume responsibility throughout the supply

chain to guarantee fuel quality in order to protect

their public image and market share. (See Interna-tional Experience 3 for an example.)

Smuggling is a problem in many countries. Hav-ing many ports of entry or fuels trucked in from othercountries makes it very difficult to check fuel qualitybecause of the large number of fuel samples involved.In contrast, it is relatively easy to check fuel quality atthe refinery gate, because most countries have only ahandful of refineries. Fuels that are smuggled into acountry are not certified for fuel quality, andsmuggled fuels are commonly not only cheaper butalso of inferior quality. Control of fuel smuggling is an

important element of a fuel quality policy.

Alternative FuelsAlternative fuels include gaseous fuels such as com-pressed natural gas (CNG) and liquefied petroleumgas (LPG), biofuels, and electricity. They are detailedbelow and in annex 6.

Gaseous fuelsGaseous fuels lower particulate emissions signifi-cantly compared with conventional diesel. Gaseousfuels are especially attractive in countries with abun-


dant reserves of natural gas or associated gas (fromoil production). A recent report by the EuropeanCommission on alternative fuels identified naturalgas vehicles as one of the most viable solutions to en-vironmental and energy security problems associatedwith the automotive industry (Alternative Fuels Con-tact Group 2003). For successful fuel switching, how-ever, it is necessary to consider fuel availability anddistribution networks; refueling infrastructure; and

costs related to vehicle modification, maintenance,and operation.

A switch to gaseous fuels can be achieved eitherby conversion of existing vehicles running on liquidfuels or by purchase of new vehicles manufactured touse gaseous fuels. Vehicles can be made to run onboth liquid and gaseous fuels or only on a gaseousfuel. Engines converted or manufactured to operateon a single fuel (CNG or LPG) can be optimized for

Market-Based Approach to Tackling Abuses in Fuel Markets:“Pure for Sure” in India

Bharat Petroleum in India has recognized that one of the basic needs of fuel-buying customers is standard-com-

pliant quality and an accurate quantity of fuel sold. As one of the major initiatives in this direction, Bharat Petro-

leum has launched a nationwide voluntary effort to dispense standard-compliant quality and the correct quantity

of fuel. This program is being launched in phases. The first phase consists of certifying retail outlets covering 108

sites in six of the country’s largest cities. Another 600 to 700 sites are to follow, in two more phases in other cities.

The retail outlets covered under this program display the “Pure for Sure” sign. At such retail outlets, Bharat Petro-

leum guarantees that the correct quality and quantity are dispensed. In order to enable them to do so, strict qual-

ity control and tracking measures have been put in place from the supply point (depot) to the customer’s fuel tank.

Before certification, the retail outlets are subjected to stringent tests to ensure that all parameters of the program

are strictly adhered to. The main pillars of this program are:

� Tamper-proof locks. Products are supplied to retail outlets in modified tank lorries fitted with tamper-proof

locks.

� Comprehensive sealing. Dispensing units are sealed in a comprehensive manner, which makes meter tam-

pering impossible.

� Periodic and surprise checks by staff. Stringent periodic and surprise checks are carried out to check for

correct delivery and the sealing of the pumps.

� Testing of product samples. Regular comprehensive testing of samples is done without warning.

� Certification of retail outlets. Periodic audits and recertification of retail outlets are carried out by a reputed

certification agency.

� Dedicated staff. Dedicated field staff has been placed to monitor and sustain the program. Anonymously con-

ducted audits and extensive inspections are carried out at these retail outlets to ensure that they continue to

comply with the requirements of the program.

These measures have been supported by the Mashelkar Committee’s “Report of the Expert Committee on the

Auto Fuel Policy,” prepared for the government of India, which has recommended that similar steps be made

obligatory for all petroleum companies.

Sources: Rogers 2002, Bharat Petroleum undated.




that fuel for best performance and least emissions. Ifthey have to be designed to operate on two fuels, theirperformance and emissions would be suboptimalwith regard to one or both of the fuels. From the point

of view of reducing emissions, single-(gaseous) fuel

vehicles are preferable to two-fuel vehicles.

Conversion of an existing vehicle may be cheaperfor an operator than premature replacement of a pol-luting vehicle. However, the conversion process mustbe carried out properly. Conversion of gasoline ve-hicles is much easier than that of diesel vehicles togaseous fuels. In the case of vehicles previously fu-eled by gasoline, conversion to CNG typically resultsin a power loss of about 10 percent, which can erodeconsumer acceptance. Poor conversions can also in-crease, rather than decrease, emissions of some pollut-ants other than particulate matter. In cities where am-bient ozone concentrations are high, increasedemissions of NOx, an ozone precursor, resulting fromconversion from diesel to CNG can exacerbate theozone problem. Experience with conversion of in-useheavy-duty diesel vehicles to CNG has been particu-larly problematic. It is essential that CNG conver-sions, whether stoichiometric (air-to-fuel ratio match-ing exactly the chemical requirements forcombustion) or lean burn (high air-to-fuel ratio), havea tamper-proof air-to-fuel ratio control using oxygensensors, or else emissions of either NOx or CO mayrise over time. Poor conversion can also lead to safetyhazards, as demonstrated by bus and taxi fires insome cities. It is therefore necessary for the authoritiesto develop and enforce strict regulations to control thequality of conversion. There is a particular dangerthat a large number of small operators will enter theconversion business with little or no quality control.Programs to promote fuel switching to gas need to be

implemented with a high degree of technical compe-

tence to avoid environmental and safety problems.

Because the cost of vehicles designed for gaseousfuels is higher than that of vehicles powered by con-ventional liquid fuels, partly due to the smaller size ofthe market, they will be attractive to consumers onlyif the higher incremental cost of vehicle acquisitioncan be offset through lower maintenance and opera-tional expenses. This is not the case when switching

from diesel to CNG or LPG in many countries, wherelow taxes on diesel make fuel switching from diesel toCNG or LPG financially unattractive for vehicle op-erators. In order to achieve reductions of particulate

emissions by switching to gaseous fuels, it is impor-

tant that relative tax levels are set to encourage the

substitution for diesel and not merely for gasoline.

Typical preconditions for successful implementa-tion of CNG vehicle programs include the following,and are discussed further in Frequently Asked Ques-tion 3:

� A natural gas distribution pipeline for otheruses of natural gas is in place.

� Adequate safety and performance standardsthat are monitored and enforced are established.

� A realistic plan to match supply and demandduring the initial phase is in place. Adequate re-fueling infrastructure for the number of CNGvehicles is in operation, and a sufficient numberof CNG vehicles, given the number of refuelingstations, is in existence or is soon to be estab-lished. However, dedicated CNG fleets, such asa downtown bus fleet with private refueling fa-cilities, can be a success without an extensivenetwork of refueling infrastructure in the citybecause they refuel at only a handful of locations.

� Managers of CNG fleets have made a commit-ment to provide technical support to deal withCNG-related maintenance and operationalproblems not generally encountered with dieselvehicles.

BiofuelsBiofuels are supported in a number of countries as ameans of diversifying energy sources, reducinglifecycle GHG as well as tailpipe exhaust emissions,and promoting or protecting domestic agriculture.They are promoted as a contribution to the efforts toincrease the share of renewables in world energy sup-plies as developed in the 2002 World Summit for Sus-tainable Development Plan of Implementation. Feed-stocks for biofuels, mostly plants, are much moreevenly distributed around the world than oil and gas.Two of the most commonly used biofuels are ethanol,


for which the world’s leading producers and consum-ers are Brazil (from sugarcane) and the United States(from corn), and biodiesel, for which the world’sleader is the Federal Republic of Germany (from rape-

seed oil). Biodiesel fuels are oxygenated organic com-pounds—methyl or ethyl esters—derived from a vari-ety of renewable sources such as vegetable oil, animalfat, and cooking oil. The technologies for making

Since CNG produces markedly lower particulate emissions than diesel, why not promoteswitching from diesel to CNG in all cities with serious particulate air pollution?

F A Q 3

There are several important considerations in deciding whether switch-

ing from diesel to CNG is a viable and cost-effective policy option for air

quality improvement.

� Do diesel emissions contribute significantly to ambient particu-

late pollution in the city? If so, then switching to gas could produce

significant improvements in air quality.

� Are there sufficient supplies of natural gas that are available to

the city? If yes, CNG is a potentially viable option. Having large do-

mestic reserves of gas alone is not sufficient for a successful CNG

vehicle program. Constructing gas distribution pipelines is expen-

sive, so that large industrial users of natural gas need to be in close

proximity to justify a pipeline network. If there are abundant domestic

gas supplies and an adequate gas distribution network in place,

natural gas may be an attractive option for use in the transport sec-

tor even if diesel combustion does not contribute significantly to air

pollution, in order to diversify energy sources. If the country is short

on natural gas and is importing liquefied natural gas, then CNG is

unlikely to be economic compared with diesel.

� Is diesel taxed sufficiently? In most developing countries, diesel is

not taxed much and is typically taxed much less than gasoline. Un-

der these circumstances, substantial subsidies would be needed to

make switching from diesel to CNG close to financially neutral. If

CNG for all vehicle users is made cheap compared with diesel, the

first result would be mass conversion of gasoline to CNG, not diesel

to CNG, resulting in a significant loss of fuel tax revenue to the gov-

ernment. In Argentina, the largest CNG vehicle market in the world,

there has been virtually no fuel switching from diesel to CNG. In

Delhi, India, where CNG conversion was mandated for certain com-

mercial diesel vehicles (buses and taxis), large-scale conversion

from gasoline to CNG among private cars created a shortage of

CNG in the early days of the conversion program.

�� Are the potentially higher costs of CNG vehicle maintenance

and the need for suitably trained technical staff taken into ac-

count in assessing conversion from diesel to CNG? There is

ample evidence from around the world that diesel vehicles are much

more robust than CNG vehicles. The evidence is especially strong in

the case of buses, which are otherwise ideally suited for alternative

fuels because buses in large fleets all “come home” to one refueling

station at night, making fueling with CNG easier. An extensive pro-

gram carried out by the bus transit authority in New York City found

that CNG buses are the most technically challenging of the three

technologies tested in terms of operation and maintenance: CNG,

diesel-electric hybrid, and “clean diesel” technology using particulate

filters and ultralow sulfur diesel (MTA New York City Transit Depart-

ment of Buses 2000). In a recent review of natural gas bus fleet ex-

perience in the United States conducted by the National Renewable

Energy Laboratory (NREL), the majority of fleets participating in the

survey reported higher overall costs for the natural gas buses. In the

same survey, the most critical factor cited for a successful natural

gas program was training, especially of maintenance personnel

(Eudy 2002). As with any program, experience varies from location

to location and bus operator to bus operator; one fleet operator in the

NREL study reported no failures, only successes, and extreme satis-

faction with natural gas buses. As natural gas vehicle technology

matures, many of the maintenance difficulties and even higher first

costs are expected to diminish. The key is to understand and antici-

pate potential technical difficulties and incremental costs, and have

a total commitment at all levels of the organization for the long-term

sustainability of CNG vehicle programs.

�� Is fundamental reform in the transport sector needed to make

the operation of vehicles being targeted financially viable? Re-

form issues, such as in a financially nonviable bus sector, may need

to be dealt with first before saddling cash-strapped fleet operators

with financially and technically challenging fuel switching from diesel

to CNG.

In short, in countries with an abundant domestic supply of natural gas

and a reasonable fiscal policy toward diesel, fuel switching from diesel

to CNG may be an excellent way of reducing air pollution. In other

cases, the questions above should be carefully addressed before the

decision is made to promote fuel switching, especially if significant gov-

ernment support is needed.


ethanol from sugarcane and corn, and biodiesel fromplant- and animal-based oils, are quite mature andwell established.

The addition of ethanol, or any oxygenate for thatmatter, to gasoline reduces the emissions of CO andhydrocarbons in old engines running “rich” (low air-to-fuel ratio). Ethanol has a high octane number, andwas used extensively in Brazil as a replacement forlead. If too much ethanol is added to gasoline in a ve-hicle that has a fixed air-to-fuel ratio, however, the ve-hicle may run lean, misfire, and potentially producegreater NOx or HC emissions. The “dilution” effect ofethanol is also beneficial: denatured ethanol containslittle or no sulfur, aromatics, and olefins, so that add-ing ethanol to gasoline lowers the concentrations ofthese components that have direct and indirect harm-ful effects on health.

Ethanol-blended gasoline has little, if any, advan-tage in modern gasoline engines equipped with anoxygen sensor. The addition of 5–10 percent ethanolto gasoline raises the fuel’s volatility markedly. If fuelvolatility is not controlled, there will be significantlyhigher evaporative hydrocarbon emissions. If volatil-ity is not to be increased, hydrocarbons with four andfive carbons may have to be taken out of the basegasoline. Because they can be cheap sources of octane,this in turn adversely affects the cost of gasoline pro-duction. In cities with serious ozone problems (suchas Mexico City, Santiago, and São Paulo), the additionof ethanol to gasoline leads to higher emissions offormaldehyde, which is an ozone precursor. In re-questing a waiver from the federal oxygenate require-ment in reformulated gasoline, the California Air Re-sources Board in 2003 submitted to the U.S. EPAdocuments showing higher NOx emissions after thestate began using more ethanol-blended gasoline(Platts Oilgram Price Report 2004). Ethanol is thusmost suited for cities with little ozone problem, highconcentrations of CO, and a large stock of old-tech-nology gasoline vehicles. In most developing countrycities, outdoor CO is typically not a serious healthhazard except along heavily congested traffic corri-dors and at intersections.

Biodiesel typically has high cetane and little or nosulfur. The addition of biodiesel to conventional die-

sel can lower particulate, CO, and HC emissions.Biodiesel fuels are biodegradable, making them espe-cially suitable for marine or farm applications.Biodiesel has better lubricity than petroleum diesel,making it suitable for blending into low and ultralowsulfur diesel, which lacks lubricity. Diesel blends con-taining 5 percent or less biodiesel are widely acceptedby vehicle manufacturers. Biodiesel seems to increaseNOx emissions slightly in vehicle exhaust andbiodiesel fuels have shown poor oxidation stability.When they are used at low ambient temperatures, fil-ters may plug, and the fuel in the tank may thicken tothe point where it will not flow sufficiently for properengine operation. There are also concerns over thecompatibility of biodiesel with seals and fuel systemmaterials. Biodiesel is an excellent medium for micro-bial growth. Because water accelerates microbialgrowth and is more prevalent in biodiesel than in pe-troleum diesel, care must be taken to keep fuel tanksfree of water. An important aspect of promotingbiodiesel is to establish clearly defined and enforcedfuel quality standards, as Germany has done.

In practice, the drive for mandating the use ofbiofuels in North America and Europe in the comingyears has more to do with support of domestic agri-culture than environmental improvement. This islikely to preclude the replacement through trade ofhigh-cost, locally produced biofuels by cheaper im-ported biofuels. Diversification of energy sources,closely linked to concerns about energy security, hasalso become an increasingly important political con-sideration in recent years, especially in the UnitedStates, where environmental reasons are now scarcelymentioned in the debate on ethanol. For developingcountries there may also be agricultural and energysupply considerations favoring the promotion ofbiofuels.

The greatest barrier to widespread use of biofuels,especially biodiesel, is the higher cost of fuel produc-tion. Biofuels can be significantly more costly tomanufacture than hydrocarbon-based fuels andbiodiesel is currently much more expensive to manu-facture than ethanol. Ethanol and biodiesel alike haverequired substantial explicit and implicit subsidies inall countries that have promoted them in order to


make them competitive with petroleum gasoline anddiesel. U.S. Treasury figures show that the revenueloss from the partial exemption from the excise tax for

ethanol between fiscal year 1980 and fiscal year 2000is estimated to amount to $11 billion, adjusted to 2000US$ (U.S. GAO 2000). To date, the policy of the gov-

Biofuels are renewable and hence should surely form an important component of sustainabletransport, so shouldn’t all governments actively promote biofuels?

F A Q 4

Biofuels such as ethanol and biodiesel are actively promoted in North

America and the EU. The EU has in fact set time-bound targets for in-

creasing the use of biofuels in transportation. The European Commis-

sion acknowledges that biofuels will offer little emission advantage over

gasoline and diesel in the future given the extremely stringent emission

standards that will come into effect in the coming years, and that

biodiesel costs two to three times as much to produce as petroleum-de-

rived diesel (European Commission 2001). The largest European envi-

ronmental citizens’ organization, the European Environmental Bureau,

has issued statements rejecting the promotion of biofuels derived from

conventional agricultural crops on the grounds that there are much

cheaper options for reducing greenhouse gas emissions, and that in-

tensive cultivation of agricultural crops can worsen ecosystem health

and reduce biodiversity (Jonk 2002).

For developing countries, biofuels offer greater emission reduction ad-

vantages than in industrial countries. That said, most developing coun-

tries embarking on biofuel projects will need to provide considerable fi-

nancial assistance to make biofuels competitive with conventional fuels

and even other alternative fuels. Under what circumstances would it

make sense for developing countries to promote biofuels?

� Is the cost of procuring petroleum products high? Landlocked

countries, or remote parts of large countries, with no sources of pe-

troleum products nearby are probably already paying very high

prices for gasoline and diesel fuel. High landed costs of gasoline

and diesel can therefore help to offset the relatively high production

costs of biofuels, making the latter more competitive. Conversely,

countries with easy access to competitive international oil markets,

and with relatively well-established internal distribution lines, will find

it more difficult to promote biofuels from an economic perspective.

� Are suitable biofuel feedstocks available in sufficient quantities

that can be collected and transported to a processing plant at

reasonable cost? The conditions for biofuel promotion are obvi-

ously much more favorable if there are already surplus crops or

products (such as molasses) that are suitable feedstocks for biofuel

production, than if land must be cleared and planted with new en-

ergy crops. The issue is whether the surplus crops or products are

concentrated within a small radius so that they can be collected in

sufficient quantity and transported to a processing plant at reason-

able cost.

� Are taxes on petroleum fuels sufficiently high to allow partial or

full tax exemption to make biofuels competitive? The reason

Germany is able to provide a considerable tax benefit to biodiesel is

that petroleum diesel carries a very high tax rate. Assuming the cost

of biofuel production is higher than the landed costs of petroleum fu-

els, tax exemption—an implicit subsidy—is typically the first line of

approach for government assistance.

� If considerable subsidies are needed, what is the objective of

the biofuel program? While biofuels have been marketed for de-

cades, no country to date has been able to sustain a biofuel pro-

gram without considerable implicit and explicit subsidies. An exami-

nation of production cost structures suggests that without significant

changes in feedstock and technologies, there will not be dramatic

cost reductions for biofuels in the foreseeable future. Therefore, if

subsidies are needed today, they are likely to be required over the

long term, and governments are likely to ask about the alternative

uses of subsidies earmarked for biofuels. If the aim is primarily to re-

duce exhaust emissions, and the tax exemption granted is on the or-

der of US$0.10–0.15 per liter as in Brazil and the United States for

ethanol, the cost-effectiveness of this approach is likely to be low.

Greenhouse gas emission reduction may also become an important

driver of biomass in general and biofuels in particular. However,

many argue that there are many cheaper options for greenhouse

gas reduction even within the transport sector, let alone in other sec-

tors. If job creation in rural communities is the goal, then the subsidies

should be compared with other government-supported job creation

schemes such as public works, or investments in infrastructure, edu-

cation, and improvement of the business environment.

In the absence of large government subsidies (explicit and implicit), the

key to a successful biofuel program is currently where there are surplus

feedstocks, especially “waste streams” from industry or agricultural pro-

cessing, and in countries where conventional liquid fuels are expensive.

In the longer term, technological breakthroughs, such as in the conver-

sion of cellulosic materials to ethanol, may greatly expand the opportu-

nities around the world for producing cost-competitive ethanol.


ernment of Brazil is to tax ethanol much less thangasoline, with the difference between gasoline andhydrous ethanol amounting to R$0.52 (aboutUS$0.18) per liter as of late 2003 (USDA 2003). Thegovernment of Germany exempts biodiesel fully fromthe mineral oil tax levied on diesel of €0.47 (aboutUS$0.57) per liter. The fiscal incentives provided tobiofuels by industrial countries need to be seen in thelight of the large subsidies provided to their agricul-tural sectors. The EU’s outlays on agriculture grew toa peak of €41 billion in 1996, excluding governmentspending on agriculture by individual member coun-tries. Grains, oilseeds, and protein crops account foralmost half of the EU’s agricultural expenditures(USDA 2002). In the United States, US$34.5 billionwas paid in subsidies to corn growers between 1995and 2002 (EWG 2003).

The main current exception to this high cost pat-tern is that of ethanol manufactured from sugarcanein Brazil, the world’s lowest-cost biofuel producer.Some industry analysts believe that the Center-Southregion of Brazil can manufacture ethanol from sugar-cane for as low as US$0.15 per liter (USDA 2003). It ispossible that this cost might be replicable in othersugarcane-producing countries if markets grew. How-ever, this low cost of production arises partly becausecapital costs form a small fraction of the total produc-tion cost due to the fact that many ethanol plants inBrazil are nearing the end of their lifecycle, and that aconsiderable amount of initial investment financingwas subsidized by the government. The replicabilityof the quoted low costs for unsubsidized new produc-ers is thus in question. The greatest promise for dra-matic cost reduction lies in ethanol production fromcellulosic materials on which there is much ongoingresearch. On balance, however, given current tech-

nologies and established production cost structures,

biofuels are unlikely to be cost-effective in combating

urban air pollution in the near to medium term.

Electric propulsionElectric vehicles are the least polluting at the point ofoperation. They also have the advantage that they al-low the recovery of some energy during braking. Forpublic transport vehicles, direct collection of current

by trams or trolleybuses is well established, with thelimitations being the inflexibility of routes and thetypically higher overall expense compared with dieselpropulsion. Electric trains or trolleybuses are there-fore most likely to be cost-effective for fixed-routepublic transport operations on high-volume routes.

For private automobiles, whose flexibility require-ments preclude the use of an external current, the fu-ture of electricity as a source of propulsion would ap-pear to be most promising with hybrid technologyrather than with battery-electric vehicles. Hybrid ve-hicles have much higher fuel economy and loweremissions than vehicles running on liquid fuels aloneand have overcome one of the biggest drawbacks tobattery-electric vehicles—vehicle range and “refuel-ing.” It may also be attractive to use hybrid technol-ogy in dedicated downtown shuttles and taxis operat-ing in sensitive areas. The development of cost-

effective hybrids appears to be the most promising

way of mobilizing the emission reduction advantages

of electric propulsion.

Vehicle TechnologyHistorically, North America, Europe, and Japan haveled the world in pursuing the best available technol-ogy for further reducing emissions from new vehicles.This has been achieved largely through regulationsrequiring vehicle manufacturers to meet extremelylow emission standards—so low that exhaust mea-surements are actually becoming difficult—and theuse of fuels to match state-of-the-art vehicle technol-ogy. A critical question for developing countries ishow quickly, and at what price, cleaner vehicle tech-nologies can be transferred to them.

Vehicle emission control technologiesVehicle emission standards are the primary technicalpolicy instrument for controlling emissions from ve-hicles in use. For gasoline vehicles, on which the useand maintenance of catalytic converters can dramati-cally reduce emissions of NOx, CO, and hydrocar-bons, the imposition and enforcement of standards

have proven a very effective environmental policy in

many countries.


Many developing countries import vehicles thatare already equipped with three-way catalytic con-verters. Once lead is removed from gasoline, the ma-jor impediment to their use disappears. In some coun-tries, rampant adulteration of gasoline with materialssuch as kerosene and poor vehicle maintenance re-sults in frequent engine misfiring and can seriouslyundermine the effectiveness of catalytic converters.After lead elimination, governments should develop a

strategy of tightening emission standards for in-use

gasoline vehicles so as to force more effective utiliza-

tion and maintenance of catalytic converters.

Aside from catalytic converters, proven controltechnologies for gasoline vehicles include positivecrankcase ventilation (recycling back exhaust gasesthat escape past the piston rings into the crankcase),canisters for controlling evaporative emissions, elec-tronic ignition, computerized ignition, and fuel injec-tion. Hydrocarbon losses during refueling can be con-trolled by returning vapor from the vehicle to theservice station tank, by using a large carbon canisteron the vehicle that traps the fuel vapors, or by adopt-ing both measures. On-board diagnostic (OBD) sys-tems that monitor emission control components arealso beginning to be well established. Proven controltechnologies for diesel vehicles include direct injec-tion, various forms of turbocharging, high-pressureinjection, electronic controls, cylinder design for highswirl, and exhaust gas recirculation (EGR). New tech-

nologies are capable of significantly reducing emis-

sions per unit of fuel consumed.

Sometimes there can be trade-offs between fueleconomy and emissions reductions, depending on thetechnology options selected. One well-known ex-ample is the inability of conventional three-way cata-lytic converters, successful in gasoline-engine applica-tions, to operate at high air-to-fuel ratios (referred toas “lean burn” and common in diesel engines). Leanburn technology increases fuel economy markedly.Three-way catalytic converters cannot work in leanburn conditions and require stoichiometric operation(the air-to-fuel ratio is set very close to the level thatwill give just enough oxygen for complete combus-tion of the fuel to CO2, with residual products of in-complete combustion reducing NOx to nitrogen).

These converters are very effective at reducing CO,HC, and NOx emissions, but at the expense of fueleconomy.

A recent example of trade-offs between fueleconomy and emissions for diesel engines is “cyclebeating” in the United States. At the heart of this inci-dent—whereby engine manufacturers were caughtselling 1.3 million engines that were emitting NOx atlevels up to three times higher than allowed by thestandards in place—was the trade-off between fueleconomy and NOx emissions. These engines had com-puter software that altered the engines’ fuel injectiontiming, improving fuel economy at the expense ofNOx emissions.

In some cases, trade-offs are not inevitable. For ex-ample, if a move to Euro II emission standards is ac-companied by a change to electronic fuel injectionand turbocharging with air-to-air aftercooling, as inEurope and with the 1991 diesel standards in theUnited States, both exhaust emissions and fuel con-sumption can be lowered. If, however, the enginemanufacturers achieve Euro II NOx emission levels byretarding injection timing and keeping mechanical in-jection systems (as in some developing countries),there is a noticeable fuel economy penalty. Possible

trade-offs between local pollutant emission rates and

fuel economy should be borne in mind in considering

alternative emission control technologies.

There can also be trade-offs between different pol-lutant control measures. The most technically chal-lenging trade-off facing engine and equipment manu-facturers in industrial countries today is between NOx

and PM emissions control for diesel engines. NOx isformed at high temperature in the presence of air, thesame conditions that reduce PM emissions. As a re-sult, it is difficult to reduce NOx and PM emissions si-multaneously using a single technical strategy. (Forbasic gasoline engine chemistry there is a trade-off be-tween CO and NOx, but this can easily be managedby use of a three-way catalyst.) Exhaust gas recircula-tion (EGR) lowers NOx emissions by reducing theamount of oxygen supplied and lowering tempera-ture during fuel combustion, but at the cost of in-creased particulate emissions. The combined use ofparticulate filters and EGR, selective catalytic reduc-


tion, or NOx adsorbers will be used by a number ofvehicle manufacturers to meet the tighter emissionstandards in the latter half of this decade in NorthAmerica and Europe. However, many of these strate-gies to reduce both PM and NOx result in decreasedfuel economy.

Technologies that carry collateral environmentaland financial benefits should be pursued as a first pri-ority. For this reason, electronic fuel control is an obvi-ous starting point for emissions improvements in allinternal combustion engines (running on liquid aswell as gaseous fuels), as it both improves fueleconomy and reduces emissions, especially in higherspeed operations. Although electronic fuel control re-quires more sophisticated maintenance, fuel cost sav-ings may quickly pay back the investment in trainingmaintenance technicians. Because of its emissions

and fuel economy benefits, electronic fuel injection

should be promoted for new vehicles.

In order to ensure the sustainability of emissionstandards, industrial countries are increasingly requir-ing vehicle manufacturers to guarantee the durabilityof exhaust control systems. In the United States, forexample, light-duty vehicles (LDVs), provided theyare properly maintained, have to meet emission stan-dards during the “useful life” of the vehicle (160,000km or higher). Higher durability requirements existfor heavier vehicles. However, vehicle manufacturerswill guarantee durability only if it can be reasonablydemonstrated that the vehicle has been well main-tained and not fueled with adulterated fuels. In partbecause maintenance and fuel adulteration problemsare common, durability requirements for the perfor-mance of emissions control devices are rare in devel-oping countries. Regulation of fuel quality and ve-

hicle maintenance through enforcement of in-use

emission standards can greatly facilitate the intro-

duction of effective durability standards for vehicle

emission systems.

Standards for new vehiclesBecause vehicle life in developing countries is oftenlonger and the vehicle population growth rate higherthan those in industrial countries, there is a strong ar-gument for mandating clean standards for new ve-

hicles. Several factors need to be taken into consider-ation when setting those standards. The most strin-gent of the standards to be implemented in the indus-trial countries (Euro IV and V for Europe and Tier 2 inthe United States—see annexes 2 and 3 for more de-tail) require sophisticated vehicle technology and se-verely reformulated fuels. Because the magnitude ofinvestments required can be substantial (for cost esti-mates of fuel reformulation, see box 2 on page 27 andannex 1), the ability to raise financing for the invest-ment projects is often a limitation. Some countriescontinue to use lead in gasoline because their domes-tic refineries have been unable to raise financing forsomething as straightforward as lead elimination. Inother countries, the inability of refineries to raise capi-tal has led to long delays in refinery revamp projectsand an increasing probability that the target dates fortighter fuel quality and emission standards will not bemet. In addition to upfront refinery investments, ve-hicle purchasers have to cover the incremental cost ofvehicles with new technology and there is also a needfor adequate investment in vehicle maintenance toreap the corresponding benefits of improved vehicletechnology and fuel quality.

Given the frequently high contribution of dieselvehicles to transport sector emissions, one option is tomove directly to the most advanced diesel technolo-gies being introduced in industrial countries. As dis-cussed previously, this may merit serious consider-ation in some cities. The assessment of whethermandating diesel particulate filters is likely to be asensible and effective policy option is discussed inFrequently Asked Question 5. This example high-lights the use of passive catalyzed particulate filters,which are the most widely used type of filters andwhich do not require an external energy input (heat,oxygen, or fuel injection). Although this example isgiven in the context of setting standards for new ve-hicles, there are also successful experiences of retrofit-ting in-use modern heavy-duty diesel engine vehicleswith particulate filters.

The considerations in Frequently Asked Question5 suggest that large and professionally managed fleetoperators in cities that have already moved to mod-ern engines, such as electronic fuel injection, are most


suited for filter technology. Industrial countries thatare moving to adopt advanced exhaust control de-vices requiring ultralow sulfur fuels have undergoneincremental tightening of standards. This has givenfleet owners and mechanics in Europe, Japan, andNorth America a number of years to move fromworking with wrenches to computerized diagnosis ofvehicles. When natural gas buses were first intro-duced in the United States, despite the sophisticatedlevel of maintenance skills available, lack of familiar-ity of mechanics with natural gas bus technology wasa major problem. In those developing countries wherefew established companies have reasonably equippedworkshops and trained mechanics, and the vast ma-jority of vehicles are serviced by roadside establish-ments with few tools, virtually no diagnostic equip-ment, little training, and limited capital, it will taketime to acquire the necessary infrastructure. Whilethere are quite a few cities that can adopt the filtertechnology effectively, and indeed some are alreadydoing so, it is not self-evident that requiring state-of-the-art emissions technology in many other develop-ing country cities for all new vehicles would be ben-eficial for the environment or cost-effective comparedto other options. It is recommended that governments

should, in consultation with vehicle users organiza-

tions, identify and enforce the most stringent stan-

dards for new vehicles that are affordable, sustain-

able, and consistent with maintenance capabilities,

and ensure that fuel quality standards are harmo-

nized with vehicle emission standards.

Regulatory loopholes are a common problem. Inone country with Euro II standards, vehicle manufac-turers are allowed to put new vehicle bodies on oldengines and chassis, some of which are imported, andthese “new” vehicles do not have to meet the new ve-hicle emission standards. A problem for exhaust emis-sions as well as fuel economy in some developingcountries is the import of large, previous-generationtechnology, private cars. Controls on vehicle imports

should include those on old vehicles and close loop-

holes that can bypass stringent emission standards.

Many of the advanced vehicle technologies, par-ticularly for reducing NOx emissions of heavy dieselvehicles, are still under development. For example,

the debate on the use of NOx traps (favored by theU.S. EPA) versus selective catalytic reduction (favoredby the German federal environment authority,Umweltbundesamt) continues. There is a large fueleconomy penalty associated with NOx traps, but un-like selective catalytic reduction, their use does not re-quire addition of urea, and the U.S. EPA is concernedabout operators’ failure to fill urea storage tanks. Se-lective catalytic reduction is more durable than NOx

traps, and enables higher fuel economy. Some arguethat the concern about operators’ failure to refill ureatanks can be overcome by electronic OBD systems forNOx emissions (Global Refining & Fuels Report 2003).

It goes without saying that new emission controltechnologies need to be actually purchased and usedif they are to help improve air quality. Estimating theimpact of installing new emissions control devices onfuture vehicle purchase patterns is not alwaysstraightforward. If fleet operators perceive vehiclesmeeting new emission standards to be much morecostly to purchase, operate, and maintain—for ex-ample, if they believe that new vehicles are less robustand reliable—then they may keep existing vehicleslonger or pre-buy new vehicles ahead of the imple-mentation of new emission standards or possibly doboth. This illustrates the importance of giving ad-equate information to vehicle purchasers, providingfleet operators with prototype vehicles to test to buildtheir confidence, and working closely with fleet op-erators in the early days of new vehicle use. There areoften costs and technical problems associated withnew technology. While costs may be confidently an-ticipated to fall substantially and technical problemsovercome, developing countries for the most part are

probably well advised to let the industrial countries

bear the costs of that development, as well as the

teething problems associated with early implementa-

tion.

Many countries have no vehicle manufacture fa-cilities and no emissions certification laboratories. Forsetting emission standards for new imported vehicles,it is important to take into account the countries oforigin and the standards in these countries, includingthe driving cycles for certification. Making slightmodifications that would require recertification


would add little value to environmental improvementwhile increasing the costs of vehicle supply and en-forcement of standards.

It is also important to note that driving cycles havea significant impact on exhaust emissions. EU emis-sion standards and U.S. emission standards are notdirectly comparable because different driving cyclesare used to certify vehicles. Reference fuels used tocertify vehicles can also differ from country to coun-try. Unfortunately, there are few direct comparisonslinking different engines, fuels, and test cycles.

Motorcycles are the most widely owned motor ve-hicles in many developing countries, especially in

Asia. While four-stroke engine motorcycles can be ap-proximately 15 percent more expensive than theirtwo-stroke equivalents, they have much better fueleconomy. Especially for high annual km vehicles suchas commercial motorcycles and three-wheel taxis, ve-hicle owners may recover the extra costs over a rela-tively short period. Despite this, many individualsstill prefer to own two-stroke motorcycles or three-wheel taxis because they are easier to keep on theroad with self-maintenance. This is a classic case of aconflict between the private interests of vehicle own-ers and the public interest of controlling vehicularemissions. Because of the narrow and falling cost dif-

(continued)

When would it make sense to install passive catalyzed particulate filters?F A Q 5

Passive catalyzed particulate filters are capable of reducing particulate

emissions to 0.01–0.02 g/kWh. Whether adopting them is a sensible

step to take depends on a number of factors. For the program to be rea-

sonably successful, the following conditions need to be met.

� Is the supply of appropriate-quality diesel guaranteed? Diesel

fuel with consistently ultralow sulfur, below 50 wt ppm and preferably

close to 10–15 wt ppm, is the primary requirement.

� Is the duty cycle suitable for filter regeneration? Particulate filter

regeneration requires a certain minimum exhaust gas temperature

at the inlet of the filter, typically around 300°C for 20–40 percent of

the duty cycle. This is what makes filter operation in light-duty diesel

engines more difficult because the exhaust gas temperature tends to

be lower. Vehicles, even buses, operating in very congested traffic

with frequent stops and starts may encounter problems reaching this

minimum temperature and hence in filter regeneration. Some diesel

engines also operate at lower exhaust temperature than others. The

location of the filter on the vehicle can also affect performance. In

general, the filter should be located as close to the engine’s exhaust

manifold or turbocharger outlet as possible in order to minimize heat

and temperature loss between the engine and the filter. If filters must

be located far away from the engine due to space constraints, highly

efficient insulation may be required on the exhaust piping between

the engine and filter, especially in areas that experience low ambient

temperatures for part or all of the year.

� Is there a mechanism in place to allow only reputable manufac-

turers with proven technology to supply filters and their re-

placements? Particulate filter standards must be established to-

gether with enforcement and a particulate filter certification process.

Filter retrofits should be carried out only by authorized, trained, and

equipped workshops that are capable of diagnosing and repairing

previous engine conditions.

� How clean is the base engine technology? The higher the level of

engine-out emissions, the larger the filter must be in order to operate

successfully. Some engines are too “dirty” for practical application of

a particulate filter. In the United States, there has been little success

in applying catalyzed particulate filters to pre-1993 engines with cer-

tified emissions levels in excess of 0.13–0.2 g/kWh when these ve-

hicles operate in slow-speed urban duty cycles. If currently operating

vehicles are considered for retrofits, particulate filters should not be

considered in mechanically governed engines, or engines in a bad

state of repair. For these vehicles, retrofitting with more modern en-

gines, even without particulate filters, can be very effective in reduc-

ing emissions.

� Is there a culture of proactive engine maintenance? In diesel en-

gines, many engine problems result in increased black smoke from

the exhaust. While in many cases the engine will still operate, a pro-

active maintenance culture would use the black smoke as a signal

that engine maintenance was required and the engine would be re-

paired. Particulate filters mask engine problems because the black

smoke that might otherwise have indicated the problem is removed.

This excess smoke can result in clogging of the filter. This means

that more sophisticated diagnostic techniques need to be employed.

In particular, when using particulate filters, the exhaust back pres-


ferentials between the two- and four-stroke technolo-gies, the high probability that poor maintenance willfurther accentuate the emissions of two-strokes, andthe undisputedly high pollution caused by in-usetwo-strokes, the most effective pathway for address-ing motorcycle emissions may be to move to four-stroke engine technology. Emerging technologies mayenable two-stroke engine vehicles to meet very tightemission standards so that two-stroke engines are notinherently polluting, but the important policy ques-tion is what emission standards will be set for newtwo- and three-wheel vehicles. Standards that arevery lenient and easily allow the continuing use of

traditional two-stroke engines, or worse, differenti-ated standards that are more lenient for two-strokethan four-stroke engines, should be re-examined andtightened. Developing countries would be well ad-

vised to require suitably tight emission standards

achievable by four-stroke engines for all except the

very smallest (50 cubic centimeters) new motorcycles,

and for all three-wheel motorized vehicles.

Enforcing policy for in-use vehiclesEstablishing fuel- and vehicle-emission standards isan important first step, but such standards need to beeffectively enforced on vehicles in use. For controlling

When would it make sense to install passive catalyzed particulate filters? (continued)

sure should be continuously monitored and checked, and remedial

action taken immediately if there are signs of the back pressure

building up.

� Are there suitably trained mechanics and appropriate equip-

ment to deal with mechanical problems? If maintenance and re-

pair personnel are not well versed in dealing with computer diagnos-

tics and electronic fuel injection, and equipped with appropriate

diagnosis and repair tools, filter operation may present a major chal-

lenge. A good and adequately funded maintenance program is a

precondition for successful filter operation. Such a requirement is not

specific to particulate filters or even to advanced diesel engine and

exhaust control technology in general, but applies equally to “very

clean” gasoline engines or CNG engines.

� Will fleet operators be able to recover the incremental cost? If

bus fares are controlled by the government and difficult to raise,

moving to more expensive technology and fuel could pose financial

problems for fleet operators.

� Will vehicles for which filters are mandatory have to compete in

the market with vehicles that are exempt from this require-

ment? Other things being equal, operators that have to install fil-

ters will be carrying higher operating costs, making them less

competitive if there are differential regulations for different fleets.

Careful consideration may have to be given to pricing and taxa-

tion policies if the move to cleaner technology reduces the (often

limited) profit margin for some and not for others, or else vehicle

owners will resist particulate filters (or any other technology for

reducing emissions).

Because of the above requirements, moving to particulate filter technol-

ogy will have higher chances of success with large fleet operators than

small operators owning only one or two vehicles. Whether particulate fil-

ters will bring marked relief from urban air pollution depends on several

factors.

� What are the current particulate emission standards? If the

standards are already at the equivalent of Euro II or the 1993

U.S. standards for new vehicles, then going down to 0.01–0.02 g/

kWh for a large number of heavy-duty diesel vehicles may make

a noticeable difference in air quality. If recent models and all in-

use heavy-duty diesel vehicles are emitting close to 1 g/kWh or

higher, then going down to 0.1 g/kWh—which is possible with

sulfur diesel of 500 wt ppm and upgraded engine technology—

may still achieve significant air quality improvement at much less

cost.

� Is the incremental cost likely to be of the magnitude that

could slow down vehicle renewal markedly? The incremental

cost of going to vehicle technology that includes particulate fil-

ters may be substantial if fleet operators are spending very little

on vehicle maintenance and repair at present because vehicle

technology is relatively simple, problems are easy to diagnose

(because most vehicle parts are mechanically controlled), and

repairs can be performed in-house without having to buy expen-

sive parts—that is, if the predominant culture is self-diagnosis

and self-repair. Under these circumstances, operators may “pre-

buy” new vehicles before filters become mandatory, and keep old

vehicles longer than they would have otherwise.

F A Q 5


emissions from in-use vehicles by means of proper ve-hicle maintenance, a number of governments have in-troduced inspection and maintenance (I/M) pro-grams (see annex 6). The underlying principle of

inspection and maintenance programs is to identify

vehicles that are not in compliance with emission

standards and to get them repaired or replaced.

Where vehicles grossly exceed emission standardsand where it may not be cost-effective to repair ve-hicles to bring the emissions down to the maximumlevels allowed, incentives may be offered to secure theimmediate withdrawal of such vehicles from opera-tion. Three main types of incentive have been used totry to improve the average quality of vehicle fleets: (1)incentives to scrap without replacement (cash for

scrappage); (2) incentives to replace with new, or lesspolluting, vehicles (cash for replacement); and (3) otherfiscal and administrative devices that, although notoffering direct financial incentives to scrap, operatethrough their impact on the scrapping decision (indi-

rect scrappage incentives). The design of scrappageschemes is discussed in the next section. They are noteasy to implement. However, if carefully designed to

ensure cost-effectiveness, vehicle replacement

schemes can be effective in reducing emissions from

in-use vehicles.

The success of emission standards for in-use ve-hicles in industrial countries has been based on thesetting of realistic standards and wide acceptance ofthe need to comply. Vehicle fleets in many developingcountries, in contrast, consist predominantly of older,earlier-generation vehicles with correspondingly highemissions. Lack of regular preventive maintenanceand proper tuning increases emission levels further.Poor maintenance of gasoline-fueled vehicles can in-crease CO emissions by two orders of magnitude. In-correct injection timing in diesel engines can increaseNOx emissions three-fold. A plugged air filter can in-crease particulate and CO emissions in both gasolineand diesel engines. Turbochargers and injectors indiesel trucks that are not replaced when they shouldbe because these replacements are costly for theowner increase emissions. Old, poorly maintained ve-hicles may have leakage past rings and crankcasevents open to the atmosphere, increasing crankcase

emissions. These considerations show the vital im-

portance to air quality of improved maintenance

practices and operation of in-service vehicles.

High costs of compliance raise a collateral prob-lem in the case of emission standards for in-use ve-hicles. Setting standards that are almost sure to resultin high rates of noncompliance is bound to lead toevasion, corruption, “false passes” in inspections, andthe public perception of environmental regulations asfundamentally flawed. A more effective policy mightbe to set relatively lenient, but rigorously enforced,standards (perhaps expecting a “real” failure rate ofabout 20 percent). Standards could then be tightenedprogressively over time. This has been the approachadopted in Mexico City. For this approach, it is neces-sary to sample vehicles and measure emissions to ob-tain a distribution profile of emissions before finaliz-ing the emission standards. It is advisable to set

emission standards for in-use vehicles that are

achievable, with some effort, by a majority of vehicles,

but that effectively eliminate the worst polluters.

Emission standards for in-use vehicles can be en-forced only if emission levels are accurately mea-sured, gross polluters are correctly identified, andthe vehicles found to be in violation are properlyrepaired to lower emissions. I/M programs in de-veloping countries must meet these key require-ments for successful implementation (for more de-tail, see annex 7):

A vehicle emissions inspection system using centralized, high-vol-

ume, test-only centers operated by private sector firms, as shown

here in Mexico City, can be effective.


� An up-to-date and accurate vehicle registrationrecord is a first prerequisite for identifying grosspolluters and getting them to the inspection cen-ters. A requirement to display a visible stickercertifying that the vehicle has been inspectedand passed, under penalty of a fine largeenough to deter evasion, is a common approachfor tackling this problem. An accurate vehicle

registration system can help ensure that target

gross-polluting vehicles are made to show up at

test centers.

� The emissions inspection system must be cred-ible. If gross polluters pass the test because theirmeasured emission levels are low, and if cleanvehicles fail because they are falsely identifiedas gross polluters, then the credibility of the sys-tem is called into question and public accep-tance falls. Avoiding this situation requireschoosing a test protocol that is difficult to cheaton or to bypass, and implementing rigorous au-dit and supervision schemes. It also requireswell-trained technical staff and proper equip-ment. The lack of proper equipment mainte-nance, inadequate calibration, and unsuitabletest procedures are three common problemsleading to inaccurate measurements. Tamperingwith test results by I/M staff for financial gain isanother common problem. Independent audit-ing of established test procedures is required.Emission levels need to be accurately measured,

and vehicle testing procedures devised to dis-

courage temporary, “pass-the-test” adjustment

of engine settings.

� There must be proper follow-up once gross pol-luters are correctly identified. This requires ad-equate infrastructure for vehicle servicing in-cluding good diagnostic equipment andqualified technicians. In the absence of effectiveenforcement of emission standards, the marketfor service and repair facilities is typically lim-ited, focusing on the vehicle’s drivability but notthat it be repaired to be less polluting. Underthese circumstances, much cheaper and inferiorcounterfeit spare parts are often widely used in

vehicle repair. As the authorities start enforcingstandards, given the right policy framework, themarket will respond by expanding service andrepair facilities. Identified gross polluters need

to be repaired or scrapped.

Probably the most extensive experience in devel-oping an effective I/M program in a developingcountry is that of Mexico City (see annex 7). Many ofthe lessons learned in the evolution of I/M in MexicoCity, particularly in respect to the use of centralized,high-volume, test-only centers operating in the handsof the private sector and market incentives to secureeffective testing, are applicable to other developingcountry cities (Kojima and Bacon 2001b). It is note-worthy that in Mexico City in January 1996, despitethe political implications, licenses were withdrawnfrom all the 600 decentralized test-and-repair garagesagainst the backdrop of an estimated 50 percent of ap-proval certificates being falsely obtained and increas-ing public opinion of the emissions control programas being fundamentally fraudulent.

For a privatized I/M system to be effective, thefollowing conditions have to be met:

� The government must be willing and able toprovide the resources for auditing and supervis-ing the program (even if the supervision isoutsourced) that are needed to guarantee its ob-jectivity and transparency. This includes remote-auditing and developing centralized softwarethat controls not only data entry, storage, real-time transmission of the captured data to a cen-tral database, and analysis, but also equipmentcalibration, zero-referencing, test protocols, anddigital signature and printing of the certificate.

� An up-to-date and reliable vehicle registrationsystem needs to be in place, with retirement ofold vehicles accurately reflected.

� The legal framework must include sanctions forfailure to carry out the testing protocols correctly.

� Testing stations must be subject to monitoringby independent bodies. All testing centers mustbe subject to equally rigorous implementation ofprotocols and inspection of their procedures.


� Testing protocols should enable reproducible re-sults to be obtained across different test centerswith a large numbers of operators and instru-ments (possibly of different makes), produce ac-curate and meaningful results, and minimizethe chances that individual testers will give falsepasses.

� The testing technology has to be able to preventthe use of temporary “tuning” that enables a ve-hicle to pass the test even though it cannot sus-tain that level of performance for regular driv-ing.

� Display of the test certificate on vehicles mustbe efficiently monitored.

� The fine for not displaying or not having a legalemissions test certificate must be sufficientlyhigh to act as an incentive to pass the test.

� The number of test centers has to be limited toensure that operators of licensed centers can ob-tain a reasonable living without reducing therigor of their tests to attract customers.

It would be very difficult, if not impossible inpractice, for one government agency to apply sanc-tions to another government agency in the form ofsuspension of license or firing of staff caught in cor-rupt acts. The question of who should run inspectioncenters (that is, public or private entities) is not asmuch about who is a more efficient operator as aboutwho would create a more enabling environment forsupervision and monitoring. Strong government su-

pervision and monitoring are considerably easier if

test station operators are privately operated.

Two-stroke engine lubricantsLubricants for two-stroke engines have a significantimpact on emissions because lubricants are intro-duced directly along with the intake air-fuel mixture.Of special concern are the quality and quantity of lu-bricant added to two- and three-wheel vehicles withtwo-stroke engines, which are numerous in Asia (seeannex 5). While both can be regulated, and premixedfuel supplied, enforcement of quality at the pump hashistorically been difficult to achieve. Poor lubricant

quality and addition of excess lubricant remain the

two main causes of high particulate emissions from

two-stroke engine gasoline vehicles.

The first step is to promote use of the proper lubri-cant: that is, using only 2T oil (lubricants manufac-tured for use in two-stroke engine vehicles). The useof so-called straight mineral oil, intended for use onlyin stationary engines, leads to higher emissions. Inone study testing three-wheelers commercially oper-ated as taxis in Dhaka, Bangladesh, switching fromstraight mineral oil to regular 2T oil lowered massparticulate emissions (measured in grams per km) bymore than 60 percent (annex 5, Kojima and others 2002).

The second step is to ensure that lubricant is usedonly in amounts recommended by vehicle manufac-turers. The use of excess lubricant is often practiced inthe mistaken belief that this will increase fueleconomy and enhance engine life. Lowering thequantity of lubricant added from the commonly used8 percent to 3 percent (the level recommended by thevehicle manufacturer) in the above Dhaka study low-ered particulate emissions by another 60 percent. Al-though 2T oil is more expensive than straight mineraloil, reducing the quantity added to 3 percent actuallysaves money for the drivers. Campaigns to reduce

emissions from two-stroke engines by improved lu-

brication can be inexpensive and cost-effective.

Vehicle Replacement StrategiesElimination of gross polluters can be an important in-strument for reducing transport-generated air pollu-tion because of their disproportionately high contri-butions to pollution. There is also likely to be theadded advantage of improving road safety when old,polluting vehicles are removed, since they tend to beless mechanically safe. In practice, environmentallymotivated scrappage schemes have not always beensuccessful (World Bank 2002d).

Targeting gross pollutersSome vehicles visibly pollute more than others, butsome pollute without showing visible signs. Evenamong the visible polluters, those that pollute themost cannot be determined by visual inspectionalone. Moreover, the method for identifying gross pol-


luters would need to be inexpensive and simple tocarry out. As discussed in annex 7, identification ofgross particulate emitters is extremely difficult exceptwhen tests are carried out in a sophisticated emissionslaboratory.2 Vehicle replacement strategies need to be

carefully targeted at proven high polluters if they are

to be cost-effective.

In the absence of adequate infrastructure for mea-suring emissions, or simply to save costs, vehicle ageis often used as a proxy for high emissions. However,mandatory scrappage based on vehicle age unneces-sarily penalizes those vehicle owners who have takengood care of their vehicles. Age-based scrappageschemes are especially problematic if a given vehiclecategory has a large number of owners with differentmaintenance behaviors and driving patterns. Impos-ing a relatively low age limit—beyond which vehiclesmust be taken outside the city at a minimum or evenscrapped altogether—may actually discourage thepractice of regular vehicle maintenance. If, on theother hand, the level of vehicle maintenance, as wellas accumulated kilometers, is fairly uniform across agiven vehicle category—for example, if all large busesbelong to one public sector bus company—then ve-hicle age is probably a good indicator of vehicle emis-sion levels. Even then, if the vehicles have been rundown and poorly maintained because the fleet opera-tor is cash strapped as a result of mismanagement,fare controls, or undue restrictions placed on its op-eration by government regulations, these problemsneed to be addressed in parallel. One reasonable ap-proach in countries that have been progressivelytightening emission standards is to select age limits onthe basis of when the emission standards for new ve-hicles were made more stringent. In this way, agelimit decisions are based on the dominant forms ofvehicle technology employed. The decision to impose

an age limit, as well as the age to be selected, should

take into account available data on emissions, ve-

hicle ownership patterns, past history of emission

standards for new vehicles, and the financial state of

fleet operators where fleets are involved.

Choosing a strategy: repair or scrap?After gross polluters have been identified, the nextquestion is whether they should be repaired orscrapped. A simple rule can be applied here: If the

cost of repairing a vehicle to reduce emissions to a

reasonably low level exceeds the market value of the

vehicle, then the vehicle should be scrapped.

If the decision is to repair, retrofitting with morerecent technology engines and parts, rather than sim-ply repairing to the original vehicle specifications,could be an effective strategy. Retrofitting buses un-dergoing engine overhaul in a municipal bus fleetwith much cleaner modern engines could provide agood opportunity for retrofit. An example is the Ur-ban Bus Retrofit/Rebuilt Program in the UnitedStates, which targets buses in major cities for retrofit-ting with a combination of modern engines and ex-haust-treatment systems (certified to reduce particu-late emissions) at the time of engine rebuild (U.S. EPA1999b), although these urban bus retrofits are quiteexpensive. However, less expensive versions could beconsidered: for example, an engine that is naturallyaspirated may be replaced with a similar turbo-charged block which has far lower particulate emis-sions at the same power output. Incentives for retro-fitting vehicles with catalytic converters have beenoffered in Germany and Hungary for many years. It isalso necessary to ensure the efficacy andsustainability of the proposed retrofit. For example,retrofitting catalytic converters on very old gasolinetwo-stroke engine two- and three-wheelers may notbe effective if “engine out” emission levels are high,significantly shortening the life of the converter. The

standard to which vehicles should be repaired should

be based on the cost-effectiveness of achieving differ-

ent levels of emissions.

Financial incentives may be used to encouragepremature scrapping of vehicles for environmentalreasons. These may be paid simply for the scrappingof a target vehicle or only if the target vehicle is re-placed by a less polluting one. Either way, it is impor-

2The sum of the cost of identification and the market value ofthe vehicle being scrapped or the repair cost should not exceedthe environmental benefit of removing or repairing the vehicle.While the environmental benefits are difficult to calculate pre-cisely, a scrappage program in which the cost of identifying par-ticipating vehicles exceeds their market values, for example, is un-likely to be acceptable to policymakers or the general public.


tant to ensure that the vehicles are actually scrapped,and not sold elsewhere. The cost-effectiveness of ve-hicle scrappage or replacement incentive policies de-pends on a range of variables.3 Other things beingequal, benefit per dollar spent increases as the follow-ing variables increase:

� Emission levels of vehicles scrapped� Cleanness of the replacement vehicles� Number of replacement vehicles attracted per

dollar� Residual life of the vehicles scrapped� Annual kilometers traveled of the vehicles re-

placed.

In practice these criteria are not independent ofeach other, and hence it is necessary to understandthe complexity of the vehicle and service supply mar-kets in order to ensure the cost-effectiveness of ascrappage scheme. It will usually be sensible to con-

centrate incentive schemes on high-usage, high-emis-

sion urban vehicles that still have a significant eco-

nomic life in the absence of the incentive.

Incentive schemes: the importance of the vehiclemarket structureThe effect of any incentive scheme, whether it be toscrap or to replace, depends crucially on the vehiclemarket structure. Vehicles are owned and operated byindividuals (in the case of motorcycles and cars) and

commercial operators, and both are largely motivatedby financial costs and benefits. Their response to fi-nancial incentives affects the entire vehicle market,and not just those who participate directly in scrap-page or replacement schemes. Even where a scrap-page grant is not in cash (for example, in Canada, oneoption offered was a free family public transportpass), experience throughout the world has shownthat most owners choose to replace their scrapped ve-hicles. An important question is whether the scrapped

vehicle is replaced with another vehicle, and, if so,

what effect this process has on vehicle usage and the

vehicle stock.

For car and motorcycle replacement, the marketappears to be segregated into two categories. Peoplein the first category, higher income or users of cars forbusiness, tend to replace existing vehicles with newvehicles every few years. Individuals in the secondcategory, purchasers of second vehicles or lower-in-come households, normally buy used vehicles and re-place them with younger, but still used vehicles. Eventhe bus and truck markets are often effectively segre-gated into large fleet operators at the top end of themarket, using newer vehicles, and smaller operatorssupplying that part of the market demanding lowerquality of service with older, secondhand vehicles.For example, the bus industry in Bangkok is seg-mented by vehicle quality: private sector subcontrac-tors to the publicly owned Bangkok Mass Transit Au-thority (BMTA) supply basic, low-fare, service withold, often very polluting vehicles sold to them byBMTA. Only a relatively small proportion of vehicleoperators replace old vehicles directly with new ones.Efficient design of scrappage incentive schemes de-

pends on understanding how vehicle replacement de-

cisions are taken by vehicle owners.

Cash for scrappageIn cash-for-scrappage schemes, the size of the incen-tive is an important variable. The larger the bonus, themore vehicles that are likely to be scrapped. However,if the oldest and most-polluting vehicles are the firstto respond, there will be diminishing returns from anenvironmental perspective as the scrappage bonus in-creases. A U.S. study (Hahn 1995), based on a pilot

3The total cost of a scheme is the unit cost per vehicle (C) mul-tiplied by the number of vehicles scrapped (N). The environmen-tal benefit is the difference between the product of the emissionrates per km traveled (ER) and the number of vehicle km traveled(VKT) within a given time period for the old vehicles and for theirreplacements, multiplied by the number of vehicles replaced andthe length of period over which benefits continue (L). A cost-effi-ciency indicator can thus be simply written as:

C · N

(ERold · VKTold – ERrep · VKTrep) · L · N

A more sophisticated indicator would discount benefits andcosts appearing at different points in time to a common base date.But the fundamental logic is unchanged, and this indicator can beused not only for comparing alternative scrappage schemes butalso for comparing public expenditures on scrappage schemeswith expenditures on other ways of obtaining equivalent environ-mental benefits. In principle the distance driven may not be thesame for the old and replacement vehicles: replacements may bemore comfortable and consume less fuel than the originals andhence attract greater use. But the empirical evidence is that this ef-fect is likely to be very small.


scrappage study in Delaware, concluded that a schemewith a very low incentive by U.S. standards (US$250)would likely be cost-effective, while those with higherbonuses to encourage higher take-up rates would not.The cost-effectiveness of vehicle scrappage schemesmay also fall sharply as the average environmentalquality of vehicles improves. Scrappage-without-re-

placement schemes can be effective for private cars, but

typically only if carried out on a relatively small scale

and if the vehicles to be scrapped are carefully selected.

The number of vehicles scrapped is not necessar-ily a good criterion for success. A scheme imple-mented in Norway in 1996 gave a bonus of theequivalent of US$880 (1997) for the scrapping of anyvehicle over 10 years of age and the program secureda net increase of 150,000 vehicles (above the numberthat would have been scrapped without the scheme).The scheme, however, did not impose any limits onthe replacement vehicle, many of which were alsorelatively old and polluting (ECMT 1999).4 Scrappage

schemes must be carefully designed and coordinated

with other public policies affecting the vehicle market.

Cash for replacementCash replacement schemes for scrapped vehicles of-ten insist that a new vehicle be bought, since new ve-hicles embody cleaner technology and the cleanertechnology is likely to be longer-lasting. However, at-tempts to force replacement by a new vehicle tend tobe attractive to those who are replacing relativelyyoung vehicles, such as high-income households andusers of cars for business. Experience in Denmark,France, and Italy shows that only 10 percent of annualreplacements involve replacing a car more than 10years old by a new one. A scheme in Hungary thattargeted old, highly polluting, two-stroke enginemodels and required replacement by a new modelfailed to attract many participants. It will normally

require a very large inducement indeed to bring about

the direct replacement of a significant number of very

old vehicles with new ones.

Cash-for-replacement schemes may attract ahigher take-up rate if replacement by secondhand ve-hicles also qualifies. But this has environmental ben-efits only if the emissions of the replacement vehiclesare tested and shown to be significantly lower thanthose of the vehicle being scrapped. In Greece the bo-nus was paid only if the replacement vehicle had acatalytic converter. Ideally, eligibility for a grantshould be based on actual emissions of vehiclesscrapped (and those replacing them if new vehiclepurchase is not required). This approach, however,also creates a moral hazard in that owners might in-tentionally allow their vehicles to fall into disrepair inorder to qualify. Cash-for-replacement schemes can

miss the worst polluters or, if the worst polluters are

replaced by only slightly less-polluting vehicles, can

yield marginal emission reduction benefits.

Despite these caveats, cash-for-replacementschemes can be successful if they target operatorswho normally keep their vehicles for the vehicles’whole lives, drive them regularly, and respond to eco-nomic signals and replace old vehicles (with deterio-rating fuel economy and increasing maintenancecosts) with newer, cleaner and more efficient vehicles.Large fleet operators of buses and trucks may be es-pecially responsive for this reason. Scrappage-with-

replacement schemes are most appropriate for large

commercial fleets of heavy public transport and

freight vehicles.

Indirect incentives and supporting policiesA number of other measures can supplement or sup-port scrappage programs.

The introduction and strict enforcement of envi-

ronmental standards through an I/M program is es-

sential. This approach encourages owners to keeptheir vehicles in good condition and to replace ve-hicles when it becomes increasingly expensive tomeet the standards.

Tax incentives can help. For example, the Germanand Hungarian governments give tax advantages forthe purchase of lower-pollution vehicles. A variationis a reduction in import duties for cleaner vehiclesand engines. Nepal reduced import duties on compo-nent parts for electric minibuses to replace diesel

4Aside from reduced environmental benefits, an ex-postanalysis of the program showed a poor benefit-to-cost ratio of 0.5.


equivalents that were banned in the Kathmandu Val-ley.

Public transport franchising policies can be very

important, particularly where buses are a significantpart of the problem. Both in Santiago and in Bogotá,environmental objectives have been incorporated inmunicipal public transport policies. Both schemes rec-ognized the need to maintain affordable public trans-port service. Environmentally oriented vehicle re-placement requirements were incorporated in abroader scheme that ensured the continued financialviability of the operating agencies. The Bogotá case

(see International Experience 4) illustrates that forcedscrappage of public transport vehicles can be effectivein the context of well-regulated franchise systems,even if the emission standard of new vehicles is some-what less than state-of-the-art.

In summary, the environmental benefits and costsof vehicle scrappage incentive schemes should becompared with other fiscal and administrative instru-ments. Hence, scrappage incentive schemes should be

designed as part of a general strategy and supported,

as necessary, by other administrative and fiscal inter-

ventions.

Vehicle Replacement in Bogotá, Colombia, and Delhi, India

In preparing to implement TransMilenio, Bogotá’s bus rapid transit system, the government of Bogotá took a series of mea-

sures that included the replacement of old buses by new. More specifically, the government:

� Required that operators bringing in new buses licensed to operate in TransMilenio demonstrate that the equivalent number

of old buses were scrapped and their licenses canceled. For each new articulated bus operating in TransMilenio, 3.6 execu-

tive buses, 2.7 regular buses, 5.3 medium-size buses, or 10.7 microbuses had to be scrapped. The operators were required

to produce a document certified by an international auditor proving that old buses had been scrapped.

� Suspended, in the early 1990s, registration of additional public transport vehicles for two and a half years (later extended to

31 December 2000, when TransMilenio became operational). However, the bus population continued to increase rapidly due

to illegal entry.

In Delhi, India, some 60,000 commercial three-wheel auto-rickshaws have been replaced. This was achieved in two phases.

The first phase involved the implementation of a Delhi Government program (which the Supreme Court intervened to expe-

dite) to replace vehicles older than 15 years by new ones complying with the latest emission standards. Between July 1998

and March 2000, nearly 21,000 auto-rickshaws older than 15 years (constituting one-fourth of the auto-rickshaw population)

were replaced with those meeting the Indian 1996 emission standards (the prevalent standards for new vehicles at that time).

The second phase followed two Supreme Court mandates that called for: (1) the replacement of all pre-1990 auto-rickshaws

and taxis with new vehicles on clean fuels, and (2) financial incentives for the replacement of all post-1990 auto-rickshaws

and taxis with new vehicles on clean fuels. Although the court required financial incentives only for the second of its orders,

the Delhi Government extended the package of incentives for the replacement of pre-1990 auto-rickshaws. Implementation of

these orders started in May 2000, and more than 38,000 auto-rickshaws of pre-1991 vintage have been replaced with new

ones powered by four-stroke CNG engines to date. In both cases, incentives combined with strong enforcement were em-

ployed.

Both phases of the Delhi program were successful because the vehicle owners saw a net benefit. In the first phase, the own-

ers could get rid of very old vehicles and purchase new, superior products at an effectively discounted price. In the second

phase, the higher initial cost of the replacement vehicle was offset by the lower fuel cost. Interestingly, in both phases, most

owners preferred to scrap old auto-rickshaws even though they were given the option of selling them to buyers outside of

Delhi.

Sources: World Bank 2002e, Iyer 2003.



53

Reducing Fuel Consumption perUnit of Movement

Reduced fuel consumption per unit of movement canhave direct benefits for reducing the emissions ofharmful local pollutants such as fine particulates, andcan also bring reductions in greenhouse gas (GHG)emissions.1 The main exceptions to this occur whenmeasures to improve fuel economy are made at theexpense of increased emissions of local pollutants (seediscussion of “lean burn” in chapter 3, Vehicle Tech-nology) or by shifting from gasoline to diesel.

There are several ways in which fuel consumptioncan be reduced. One way is to increase the inherentfuel economy of individual vehicles. Another is to en-courage vehicle operation that minimizes fuel con-sumption. A well-known approach is to smooth outtraffic speed so that for the same distance traveled,much less fuel is used. Shifting the mode of transport,such as from motorized to nonmotorized transport orfrom private cars to public transport, can also reducefuel consumption. Promoting or protectingnonmotorized transport should generally result insignificantly reduced fuel consumption and should besupported for equity reasons in any case.

Improving Fuel Efficiency throughVehicle TechnologyVehicle technology options for improving fueleconomy include increasing engine efficiency, de-creasing vehicle weight, improving aerodynamics,and lowering rolling friction for tires. Fuel efficiencyincreases with increasing engine compression ratios. Italso increases at a higher air-to-fuel ratio (lean burn),

with precise injection timing, and anything else thatincreases the completeness of fuel combustion.Switching from mechanical to electronic fuel injec-

tion is one of the most important steps developing

countries can take to improve fuel economy.

Everything else remaining the same, lowering thevehicle weight (and hence decreasing the vehicle size)and power increases fuel economy. The principal rea-son fuel economy in the mainstream vehicle fleet insome industrial countries has not improved over thepast two decades is that technological advances in en-ergy efficiency have been overwhelmed by the trendtowards increased power and speed of vehicles andthe simultaneous shift to larger and heavier vehicles.In many developing countries, vehicle size and powerare much smaller than in industrial countries, so thatthere is not much scope to reduce them further. Over-all reductions in fuel consumption and emissions arepossible in industrial countries and in developingcountries in the future if motorists can be persuaded,and encouraged through fiscal policies, to accept re-duced engine power and smaller vehicle size. Smallvehicles must also be crashworthy, to ensure thatthere is not a serious trade-off between safety and fueleconomy. Standards for new vehicles must be con-

tinuously reviewed and revised to take advantage of

the technological potential to improve fuel economy

and reduce air pollution.

A shift from gasoline to diesel for fuel economyreasons can have potentially negative environmentaleffects. Diesel is an inherently more efficient fuel thangasoline, and worldwide, future demand growth formotor fuels is anticipated to favor diesel, while de-mand growth for gasoline is expected to stagnate. Un-fortunately, conventional diesel engines producemuch more particulate emissions in mass than do

1When looking at GHG emission reductions, the entire cyclefrom “well to wheel” should be examined since the consumptionof fuel is only one component of the energy costs of a vehicle.

C H A P T E R4


gasoline engines. This can be overcome by the use ofadvanced control technologies and ultralow sulfurfuel. At the present time, PM control technologies arestill emerging for light-duty diesel vehicles, and withheavy-duty vehicles, a number of preconditions mustbe met for the control devices to be effective (see Fre-quently Asked Question 5 on pages 44 and 45). Care-

ful consideration should be given to the likely impact

on particulate emissions when deciding whether to

encourage a shift from gasoline to conventional diesel

for fuel economy reasons.

Some governments have imposed fuel economystandards or negotiated them with the auto industry.The best-known case is the development of the corpo-rate average fuel economy (CAFE) standards in theUnited States. Unfortunately, the CAFE standardshave changed little since their inception in the 1970s,with improvements in technology permitting in-creased vehicle power rather than progressively re-ducing total fuel consumption and emission levels.Low fuel taxes have contributed to this trend. Thelack of progress in U.S. fuel economy has been exacer-bated by legislation that sets lower fuel economy andemission standards for light trucks—such as pick-uptrucks, passenger vans, and sport utility vehicles(SUVs)—and by a large tax exemption on these ve-hicles for small businesses. As a result, the averagefuel economy of all new vehicles in the United Statespeaked in the mid-1980s, and has declined since. Theincreasing fraction of light trucks in the total new ve-hicle sales, now exceeding 50 percent, has been themain cause. For light trucks with low fuel economy,there is no comparable “gas guzzler tax,” a graduatedtax on new passenger cars based on fuel economy.Some of the largest SUVs are in fact classified asheavy-duty trucks. There have been cases where ve-hicle manufacturers increased weight ratings of somepopular pick-up models to avoid fuel economy stan-dards, or redesigned them to avoid being classified aslight trucks (Wells 2003). It is important to minimize

opportunities for vehicle manufacturers to exploit

loopholes or act in an opportunistic manner in an at-

tempt to circumvent fuel economy and emission stan-

dards.

In the EU, where fuel taxation is generally muchhigher, vehicle manufacturers entered voluntaryagreements with the European Commission to reduceaverage CO2 emissions from new passenger cars to140 g/km by 2008 and to 120 g/km by 2012. TheCommission confirmed at the end of 2002 that Euro-pean manufacturers were on course to meet their tar-get. However, by the end of 2003, available data indi-cated that European automotive manufacturers areunlikely to meet 2012 goals for CO2 emission reduc-tions. Monitoring figures showed that emissions in-creased from 164 g/km in 2001 to 165 g/km in 2002.As in the United States, this is primarily because ofthe trend towards larger and heavier vehicles with anemphasis on performance in terms of acceleration andtop speed rather than fuel economy. The “perfor-mance race” of recent years lies at the heart of theproblem. Manufacturers cited growing consumer de-mand for SUVs and higher vehicle weights associatedwith tighter safety standards as two primary reasonsfor this trend (Automotive Environment Analyst2003a).

Among developing countries, the People’s Repub-lic of China is planning to phase in minimum fueleconomy standards on new cars beginning in July2005. The proposed standards are reportedly muchmore stringent than those in the United States (Auto-motive Environment Analyst 2003b).

Lastly, it is important to bear in mind that traveldemand has a tendency to increase as a result of atechnical improvement in energy efficiency, and this isknown as the rebound effect. As with traffic manage-

ment discussed below, it is important to offset this

tendency for increased travel resulting from improved

fuel economy by the simultaneous introduction of de-

mand management instruments.

Increasing Fuel Efficiency throughVehicle OperationPoor vehicle maintenance and certain operationalpractices—such as overly retarded injection timing,not correctly inflating tires, or driving behavior char-acterized by sudden acceleration and deceleration—

CHAPTER 4. REDUCING FUEL CONSUMPTION PER UNIT OF MOVEMENT 55

lower fuel economy. Retarded injection timing in-creases fuel consumption under all circumstances.Low tire pressure increases vehicle rolling resistancecausing higher engine power levels and increasedfuel consumption by 5–10 percent, raising in parallelthe emissions of NOx and possibly particulate matter.Some studies have reported fuel consumption differ-ences of as much as 15 percent. Driver behavior alsoaffects fuel economy: minimizing unnecessary brak-ing, observing the speed limit, and avoiding exces-sively rapid acceleration can improve fuel economyby a few percent over normal driving behavior. It ispossible to increase fuel economy by another few per-cent via optimal vehicle maintenance. Poorly main-tained roads make it difficult for drivers to maintain asteady speed and lower fuel economy markedly. Fur-ther, if roads are in very bad condition, there may bean incentive for larger four-wheel drive vehicles orSUVs, even in situations where a small car might oth-erwise do. Overloading, a common practice in manydeveloping countries, accelerates damage to roads.Proper vehicle and road maintenance as well as

proper vehicle operation can improve fuel economy

significantly.

In countries where transportation fuels are cheap,as in Turkmenistan or Venezuela, a recent report bythe U.S. Congressional Budget Office on instrumentsfor improved fuel economy may be informative. Ex-amining three different approaches to decreasing fuelconsumption by 10 percent, the report indicated thatthe cheapest and most effective path would be a sub-stantial increase in the fuel tax. Simply raising theCAFE standard would be the most costly to consum-ers, adding an average US$153 to the cost of a newvehicle. Raising gasoline taxes would not only costless than the other two approaches considered (bothof which involved raising CAFE standards), but itwould start reducing consumption immediately, andthe market effect would gradually drive the transitionto more fuel-efficient vehicles (Automotive Environ-ment Analyst 2004). Raising fuel prices to reflect the

real cost to the economy is an important consider-

ation in stimulating fuel economy as well as in reduc-

ing non-essential trips, especially in developing coun-

tries.

Encouraging Nonmotorized TransportTraffic mix is an important determinant of emissionlevels for two reasons. First, because most emissionsfrom motorized vehicles are highest at cold start ofthe engine, short trips are disproportionately pollut-ing. These are the trips that are most suitable fornonmotorized transport (NMT). Unfortunately, inmany developing country cities, walking or usingother nonmotorized forms of transport is so inconve-nient and dangerous that even very short motorizedtrips are common. Eliminating impediments to NMTby providing adequate sidewalks and bicycle lanesand ensuring the safety of pedestrians and cyclistscan deter the use of the most polluting motorized ve-hicles for short trips. An equally important, or aneven more important, consideration is that segregatedlanes for bicyclists and pedestrians can enhance pub-lic safety and help to reduce the number of deathsand injuries markedly—every year more than 1 mil-lion die in road accidents worldwide, over four-fifthsin developing countries, and as many as 50 millionare injured (The Economist 2004). Provision for safe

and comfortable walking and other forms of

nonmotorized transport should be an integral part of

an urban air quality strategy.

Second, traffic mix has a substantial impact onvariability of traffic speed. This is a serious problemwhere motorized and nonmotorized traffic share roadspace. Measures to segregate these types of traffic onmain thoroughfares are thus as important for environ-mental as for safety and efficiency reasons. For ex-ample, in Dhaka, the presence of a large number ofcycle rickshaws has historically made the operation ofbus transport very difficult. In contrast, in residentialareas it may be better to use traffic-calming measuresto harmonize speeds of different traffic categories at asafe level. Traffic calming also has the indirect effect ofdeterring traffic from using residential roads as ashort cut. Careful differentiation of traffic segregation

policies by type of road can reduce environmental im-

pacts and accidents while increasing average speed

for all traffic.2

2See the various reports of the U.K. Standing Advisory Com-mittee for Trunk Road Assessment.


Bicycles can account for as much as 50 percent oftotal movements in some low-income countries in Af-rica and in many Asian cities. Despite this, the bicycletends either to be neglected or is actively discrimi-nated against. This is partly because the bicycle isconsidered by government to be associated with pov-erty, and hence to be a mode that will disappear as in-comes increase. In China, bicycles are viewed by gov-ernment as a barrier to modern road transport andare consequently being banned from roadways inmany cities. Often this is being done without a conse-quent provision of alternative space for bicycles, orwithout adequate consideration to mobility decisionsof displaced bicyclists. In Vietnam, it is striking howquickly bicycles are being replaced by (predominantlyfour-stroke) motorcycles as incomes rise in the majorcities, and consequently how high is the level of indi-vidual mobility in those cities. There is a danger that

the motorcycle is the first step in the direction of reli-

ance on the private automobile, which would not be

sustainable given the density of the central cities.

The appropriate policy response, now beingadopted in a number of industrial countries, is to

view the bicycle as an environmentally friendly modefor shorter trips and to plan positively to make it at-tractive. Some of the earlier initiatives to re-establishthe bicycle through provision of sections of bicycletrack, as in Lima, Peru, have had limited success.There are many components to making a bicycle pro-motion program work, including segregated infra-structure, provision for modal integration with publictransport, promotion (particularly through safety andsecurity campaigns), and other incentives. The mes-sage, exemplified by long experience in the Nether-lands and by the increased bicycle use that has morerecently been achieved in Bogotá, is that a comprehen-

sive package of measures, including extensive connec-

tivity in the bicycle network, is necessary in order to

make the bicycle attractive and sustainable as in-

comes rise.

Traffic managementThis report distinguishes between traffic manage-ment, which consists of supply-side measures to im-prove performance of roads with existing traffic vol-umes, and demand management, which consists ofmeasures to improve performance by reducing trafficvolumes (World Bank 2002b). The former is discussedin this chapter and the latter in the next chapter. Bothmay require some physical measures, usually referredto as “traffic engineering.” The engineering involvedin traffic management tends to have a short gestationperiod and low cost. Traffic management has the po-tential to achieve reductions in air pollution and to beaffordable, even by poor countries.

The objectives of traffic managementThe adverse impact of local air pollution is highly lo-cation-specific and, to a lesser extent, time-specific. Itis greatest where most people are exposed and whereemissions lead to high ambient concentrations (on ac-count of high emission intensity and low dispersionof pollutants). A high level of exposure is thus theproduct of a series of decisions or circumstances thatdetermine the number of trips made, their distribu-tion over space and time, the choice of routes, thedriving characteristics of drivers, and where people

Segregating bicyclists from the rest of traffic, as this road in Beijing

does, promotes nonmotorized transport, protects bicyclists’ safety,

and helps to keep traffic speed higher than when motorized and

nonmotorized traffic are mixed.


spend time. Traffic management in industrial coun-tries has been estimated to reduce emissions by 2–5percent overall, but by much greater proportions inspecific corridors or areas. Because of poor initial

traffic conditions, there is considerable potential for

traffic management to reduce fuel consumption in

many developing country cities.

Both fuel consumption and exhaust emissionsvary significantly with variability of vehicle speed.Traffic management can, in principle, reduce fuel con-sumption and exhaust emissions by making trafficflow more smoothly. Other things remaining thesame, it would therefore be desirable to manage traf-fic to secure uninterrupted movement at free-flowspeed. A steady speed is also the key to reducing theemissions of harmful pollutants per unit distancetraveled. A number of devices, such as one-way streetsystems, linked traffic signal systems, and traffic con-trol systems can contribute to smoothing traffic flow.From an environmental point of view, the most im-

portant features to address by traffic management are

the variability of traffic speed and the location of

major traffic flows, particularly congested flows.

There is one major drawback. As average trafficspeed increases so will trip lengths. Traffic manage-ment may induce more or longer trips to be made sothat congestion is little relieved and total emissionsmay even increase. Detailed evidence of the traffic-generating effects of urban ring roads has been as-sembled in analysis of the M25 motorway aroundLondon. Traffic management is likely to realize its fullpotential to reduce air pollution only if supported bymeasures to restrain new traffic generation. Hence itis important that the adverse effects of new traffic

generated by improved traffic management be offset

by the simultaneous introduction of demand manage-

ment instruments to limit traffic growth and protect

areas where exposure is greatest.

Traffic signal systemsTraffic signal control systems are the most commontraffic management instruments aiming to secure traf-fic flow objectives. However, their impact on air qual-ity has been controversial. Some have argued that be-

cause they achieve their travel flow objectives by sys-tematically bringing some traffic flows to a stop, theyare likely to increase air pollution and should be re-placed by roundabouts or fly-overs (OECD 1991).Others have challenged this claim, arguing that theimpact of traffic signals on pollution is highly situa-tion-specific.

Some conclusions on traffic signals are widely ac-cepted, however. Linking of uncoordinated signals tocreate “green waves” can reduce travel times by 10percent and emissions by a similar proportion in thecontrolled area. Allowing “near-side turn on red” (leftturn where vehicles are driven on the left side of theroad) gives another 1.5 percent improvement. Cyclelengths that minimize pollutant emissions are 50 per-cent longer than those that minimize delays, and inheavy traffic conditions these extended cycle timescan reduce emissions by up to 3 percent. The most ef-ficient systems are area traffic control (ATC) systems,which link signals across whole networks. These sys-tems can be made traffic-responsive on a real-time ba-sis but are more expensive in terms of capital equip-ment (partly because of the need for moretraffic-sensing equipment). However, in developingcountries ATC has a history of contract failure, dis-pute, and procurement difficulties, as well as opera-tional weaknesses. The Phase I ATC system inBangkok, installed in 1996, still functioned imper-fectly by 2000 because of a lack of sustained coopera-tion from the traffic police (Cracknell 2000). Effective

use of traffic signal systems to improve air quality re-

quires careful design and committed institutional co-

ordination.

Road system designRing roads and by-passes are not traffic-managementstrategies per se, but they are often advocated as thebasis on which it is possible to introduce environmen-tal traffic management. The argument is that if suchroads provide adequate capacity to navigate across atown it will be possible to keep through-traffic out ofenvironmentally sensitive areas. In some small or me-dium cities that have restricted vehicle access to cen-tral areas, this has worked well (Freiburg, Germany, is


one example). But in many areas, it has not. There aretwo main reasons for this:

� Improved radial or ring road performance in-creases the number and length of trips made tosuch an extent that total traffic and total emis-sions actually increase. This is sometimes calledthe “rebound effect.” Both average speeds andjourney times may be increasing simultaneously.

� The supporting traffic management necessary totake advantage of the “breathing space” is notimplemented. This has been a particular prob-lem in Chinese cities such as Guangzhou andShanghai.

This experience indicates that increasing infra-

structure capacity will result in improved air quality

only if embedded in a comprehensive urban transport

strategy involving parallel restraint of vehicles and

local environmental protection.

Some pollutants, such as CO, are within health-based air quality standards on average but can be ex-tremely high at urban “hot spots,” such as heavilycongested traffic corridors and intersections. CO andPM concentrations fall rapidly with increasing dis-tance from these roads. Schools, hospitals, homes forthe elderly, and shopping streets should therefore belocated several hundred meters away from busy traf-fic corridors. For existing hot spots, traffic manage-ment can be used to minimize the impact of traffic onlocal air quality. New infrastructure should be de-

signed to minimize chances of creating hot spots.

The influence of hills on emissions can be signifi-cant. Emissions do not rise substantially on dieseltrucks until the terrain is sufficiently steep to requireuse of brakes on downhills and power re-supplied onuphills. The situation is made worse if congestioncauses stop-start driving on a hill. Major arteries

through cities should avoid severe grades where possible.

Local environmental managementPedestrianization of city centers began to gain popu-larity in Europe about 40 years ago and is now a fea-ture of most city-center plans. The sort of measuresused include the shading of pathways, attractive pav-ing materials, use of materials to dissipate heat, inte-

gration with public transport stations, landscaping,and better pedestrian corridor links with major desti-nations. In contrast, pedestrians are generally poorlyserved in developing countries, where they tend to becontrolled rather than provided for. Footways are of-ten not provided; when they do exist, they are often ina poor state of repair or taken over by traders andparked vehicles. The consequence is that pedestriansare forced to walk in the highway pavement. This isnot only unsafe but also contributes to traffic conges-tion and consequently increased emissions. In manycities the roads are so unsafe for pedestrians that theyuse motorized transport even for short trips. Provi-

sion of adequate pedestrian facilities improves air

quality by keeping traffic away from sensitive, high-

exposure locations and by encouraging walking as

the preferred mode for short trips.

Other restraints on vehicle movements are usuallytargeted at particularly sensitive areas. The most com-mon spatial restrictions relate to access to central busi-ness districts (CBDs). The “cell system,” introduced inGothenburg and replicated in British towns such asOxford and Leeds, uses physical restrictions on cross-center movements to keep through-traffic of privatevehicles, but not buses, out of the CBD. Some schemesalso discriminate by vehicle type. For example, thebus franchising system in Santiago limits the numberof buses licensed to operate into the CBD, although itis not well enforced and currently under revision.Many cities such as Delhi specify particular routes forheavy goods vehicles or even ban their access to cen-tral areas during the daytime. Discriminating traffic

controls can be used to protect environmentally sen-

sitive areas.

The difficulty for many developing countries,however, is that important generators of heavy freightvehicle movements, such as ports (as in Manila) andmajor markets (as in Dhaka), are located in or close todowntown areas. Relocation of these facilities, as cur-rently being implemented in Ho Chi Minh City inVietnam, is very costly and can only be achieved overa long time period. The evidence is that integrated

planning of urban land use, urban public transport,

and traffic management is the best basis for improv-

ing air quality in the most sensitive locations.


Incident detection and intelligent transportsystemsMuch congestion in large cities can be attributed tothe dislocation effects of relatively trivial accidents.Traffic incident detection, coupled with prompt ap-propriate response, can improve traffic flow and re-duce congestion significantly. This requires appropri-ate real-time transfer of information and closecollaboration among traffic management, police,health, and rescue agencies. The ability to identify in-

cidents, remove obstructions, and redirect traffic can

help reduce traffic congestion and improve air quality.

Regulation and Control of Public RoadPassenger TransportPublic transport affects urban air pollution both di-rectly, through emissions of public transport vehicles,and indirectly, by providing an alternative to a muchlarger number of private cars (World Bank 2001). Op-eration of public transport vehicles may result inlosses of car speed, resulting in slightly higher emis-sions from cars. If public transport is not efficient, it isunlikely to contribute effectively to the reduction ofurban air pollution. However, provided that publictransport is sufficiently attractive to draw passengersaway from private vehicles to high-occupancy publictransport vehicles and is well maintained, on balancepublic transport promotion can bring significant envi-ronmental benefits (though these benefits should becalculated and not simply assumed). Policy for public

transport can minimize its direct air pollution im-

pacts by making it clean and maximize its indirect

benefits by making it sustainable and attractive.

Traffic management for public transportBuses typically move at only about two-thirds thespeed of cars because they must frequently stop andre-enter the traffic flow. Given the limited density ofbus networks, taking buses also involves longer ter-minal walking times than using private cars, with theoverall result that a bus journey usually takes at leasttwice as long as an equivalent car journey. This dis-crepancy accentuates the advantage of the private carand encourages its use. Although auto-rickshaws of-

fer point-to-point service, they also add to congestion.Mixing public transport vehicles, whether buses orauto-rickshaws, with other vehicle categories reducesthe average speed of traffic compared with whatcould be achieved if traffic were segregated. Public

transport priorities—dedicated lanes or totally segre-

gated busways—are essential to counteract the prob-

lems of mixed traffic.

The simplest measures are priority bus lanes. Butthey have major limitations. They make roadside ac-cess to premises more difficult. When they are oper-ated in the same direction as the main traffic flow,they are particularly susceptible to invasion by othertraffic. Operation against the direction of flow is moreself-enforcing but can increase pedestrian accidents.Simple non-segregated bus lanes have proven difficult

to enforce, thus providing limited improvements in

bus efficiency and air quality.

Totally segregated busways using central lanes,along with protected pedestrian crossings at stations,overcome the problems of accidents and problems ofaccess to roadside premises associated with bus lanes.Furthermore, by developing busways as trunk linksin a physically and commercially integrated network,the travel time and cost of public transport can bemade more competitive with that of the private car.Although schemes that dedicate existing road spaceto public transport may be opposed by car users, ex-

Bus lanes cannot prevent mixing public transport vehicles with other

vehicle categories, as the above photograph of a trolley bus and pri-

vate cars in Moscow shows.


perience in Curitiba, Brazil, and Bogotá (see Interna-tional Experience 5) has demonstrated that, with good

traffic management to minimize car delays, segre-

gated bus systems can produce both efficiency gains

and environmental benefits.

Improving internal efficiency of operationsThe internal efficiency of formal bus operating com-panies in developing countries can usually be im-proved by more efficient design of route networks,better cost control, and better control of performanceon the road. Some of these involve relatively modern

technology (such as automatic vehicle location) that islikely to be employed only by large, possibly area-mo-nopoly, companies. But this has to be balanced againstthe losses of efficiency inherent in monopoly opera-tion. While there are some scale economies in stafftraining, supply procurement, and management in-formation systems, international experience indicatesthat these are not of a magnitude to justify monopolyoperation of large urban systems. The advantages ofintegrated systems planning, also frequently consid-ered to justify monopoly, can be equally well achievedby separation of planning from the operation of ser-

Bus Rapid Transit in Bogotá

As part of a comprehensive urban mobility strategy including promotion of nonmotorized transportation and re-

striction of automobile use, the municipality of Bogotá has developed a bus rapid transit system called

TransMilenio. The infrastructure of the system includes exclusive busways on central lanes of major arterial roads,

roads for feeder buses, stations, and complementary facilities. Trunk line stations are closed facilities with one to

three berths varying from 40 meters (m) to 180 m in length, located in the median every 500 m on average. Trunk

lines are served by articulated diesel buses with 160-passenger capacity, while integrated feeder lines are served

by diesel buses with capacity of 80 passengers each. To maximize capacity, trunk lines accommodate express

services, which stop only at selected stations, as well as local services, which stop at all stations. This combina-

tion allows the system to carry up to 45,000 passengers per hour per direction. Services are operated by private

consortia of traditional local transport companies, associated with national and international investors and pro-

cured under competitively tendered concession contracts on a gross-cost basis. By the end of 2002, 750,000 pas-

sengers per day were carried on 41 km of exclusive lanes and 309 line km for feeders on local roads, with 61 sta-

tions, 470 articulated buses, and 241 feeder buses. During peak hours, 35,000 passengers per hour per direction

were being transported in the heaviest section. TransMilenio is intended to expand over a 15-year period to in-

clude 22 corridors with 388 km of exclusive lanes. By 2020, 85 percent of the city is envisaged to be within 500 m

of the trunk system, and the rest of the city covered by short distance feeder systems.

Overall system management is performed by a new public company (TRANSMILENIO S.A.) funded by 3 percent

of the ticket sales. TRANSMILENIO S.A. operates a control center, supervising service and passenger access.

The system was developed between January 1988 and December 2000, when service commenced. By May 2001

it carried 360,000 trips per weekday, at a ticket cost of US$0.36 and without operating subsidies on 20 km of ex-

clusive lanes, 32 stations, 162 articulated buses, and 60 feeder buses. Productivity was high, with 1,945 passen-

gers per day per bus and 325 km per day per bus. Fatalities from traffic accidents had been virtually eliminated

along the trunk corridors, and users’ travel time reduced by one-third. Measurements indicated that particulate

emissions in the corridors were reduced by up to 30 percent, simply by replacing many old, polluting vehicles with

larger, but still conventional, diesel vehicles. As the system has proved profitable for the operators, it is intended

to tighten the vehicle quality requirements in later rounds of service tenders to reduce air pollution even more.

Sources: Hidalgo 2001 and 2003.




vices, which can be competitively procured by theplanning agency. The most important requirement is

usually the need for internal incentives to efficiency,

the most effective of which is some competitive

threat.

Public transport franchisingMost public sector operations of public transport arepolitically controlled and inefficient. Yet allowingsmall informal sector operators to enter the market tosupplement or compete with the existing operator hasoften been associated with excessive supply (as inSantiago until the early 1990s), the use of old, pollut-ing vehicles (as in Lima today), or dangerous operat-ing practices (as in Delhi). Unregulated competition

can clearly be dangerous, inefficient, and environmen-

tally damaging.

But this is not inevitable. Several countries, includ-ing Denmark, Sweden, and the United Kingdom,have awarded monopoly franchises of limited dura-tion and scope on the basis of a competitively bid ten-der. This “competition for the market” allows the au-

thority to control the main policy sensitive variables,such as fares and service structures, while mobilizingcompetition to get the desired level of service at thelowest possible cost. It has shown reductions in costper bus kilometer between 20 percent and 40 percentand is now the preferred form of competition in largecities (Halcrow Fox 2000). The replacement of compe-tition “in the market” by competition “for the mar-ket” in the central area of Santiago allowed the au-thorities to get the economic benefits of competitionwithout environmental damage by simply settingminimal pollutant emission standards as a conditionfor holding any franchise, as well as by using environ-mental quality above the minimum as one of the crite-ria on which competitively tendered franchises areawarded (see International Experience 6). Institu-

tional and regulatory reform to create orderly compe-

tition for franchises has improved performance and

maintained public transport share in many countries.

For competitive tendering to be effective, a fran-chising authority must be technically and administra-tively able to design and award franchises with sen-

Addressing the Environmental Impacts of Bus Competition in Santiago, Chile

At the end of 1977, public road passenger transport in Santiago was provided by a public sector operator with

710 large (90-seat) buses, a number of strictly regulated private associations operating 3,167 regular (78-seat)

buses, and 1,558 (40-seat) “midi”-buses. The public operator lost money and service was mediocre. Between No-

vember 1979 and June 1983, both entry to the market and fares were deregulated. The public sector operator

was driven out of the market, and total capacity more than doubled. But by 1985, regular bus fares had tripled,

and the average age of buses had increased from 7 to 11.6 years. Competition concentrated on routes to the cen-

ter of the city, which became congested and polluted by buses with too-few passengers.

Initial attempts to rectify the situation included banning 20 percent of the bus fleet from operation on each day of

the week and banning buses more than 22 years old. These measures gave little relief. So, in the early 1990s, the

government introduced a system of competitive tendering for franchises to operate buses on routes entering the

city center. The capacity was thus constrained by the authorities. The fare to be offered was one main criterion in

selecting franchisees; another was the environmental characteristics of the vehicles offered. Congestion, air pollu-

tion, and fares all fell dramatically. By the mid-1990s, improved service, an important benefit of competition, had

been retained, while the drawbacks associated with competition had largely been eradicated. In a 10-year period

between 1992 and 2002, the number of buses fell from 14,000 to 7,500, and the average age of buses from 15

years to 5 years.

Sources: World Bank 2002f, López 2002.




sible environmental conditions and to monitor perfor-mance—including vehicle emissions—effectively.There is now a wealth of experience in doing this,both in industrial countries (for example, inCopenhagen and London) and developing countries(in such cities as Bogotá and Santiago). Furthermore,effective competition, either in the market or for themarket, is dependent on the commercialization or fullprivatization of the incumbent parastatal operator, asprivate operators are understandably reluctant tocompete with an agency that can rely on deficit fi-nance from its owner to ensure that it retains its posi-tion in the market. The cities that have most satisfac-

torily reconciled efficient and clean operations with

low budget burden are those that have developed ef-

fective competition.

Emission standards for public transport vehiclesA number of countries have embodied high aspira-tions for the environmental quality of public transportby enacting strict standards for public service ve-hicles. In some cities, such as Kuala Lumpur, this hashad the perverse effect of reducing the commercial vi-ability of public transport. In others, such as Bangkok,environmentally clean vehicles have been put in ser-vice only at premium fares that reduce the availabilityof service to the poor. Conversely, cities such asBogotá—which have concentrated on giving priorityto public transport vehicles but have not enforcedvery high environmental standards per se—havebeen able to reduce air pollution by attracting passen-gers into large buses and eliminating many of the old-est, most polluting, vehicles. Imposing stringent ve-

hicular emission standards without attention to the

financial sustainability of public transport opera-

tions can undermine the viability of public transport

and have counterproductive effects.

Bus priorities and bus rapid transitBus priority systems change the relative travel timesby bus and car and, particularly if supported by park-ing restraints, encourage people to use the morespace-efficient public transport modes. Congestionlevels may thus be reduced. More important, they in-crease the average speed and reduce the variability ofspeed of bus movements. For turbocharged diesel en-gines, the importance of maintaining steady engineload (steady speed or gentle speed changes) to reduceparticulate emissions cannot be overemphasized. Arange of priority measures was shown to reduce busexhaust emissions in London by 7–60 percent (table 3).

As table 3 shows, the segregated busway, or busrapid transit (BRT) system, is the most environmen-tally effective although there is minimal use of thesegregated busway in London. BRT has been devel-oped extensively as the core of mass transit systems inBogotá and Curitiba. Such systems make it possible toreplace four or five small vehicles with one larger ve-

Does privatization of public transportlead to worsening urban air pollution?

In industrial countries, publicly owned public transport operators are

subject to stringent quality standards (including for emissions) that

are satisfied adequately in many countries. But the operators usually

achieve this on the basis of substantial, and open-ended, public sub-

sidies. Where those subsidies are not available, as is now the case in

many of the independent republics that emerged from the former So-

viet Union, publicly owned buses are old, poorly maintained, and

heavily polluting. It is thus the subsidy, rather than the public owner-

ship, that allows stringent quality standards to be met.

In many countries, both industrial and developing, the size of the

subsidies has become so great that governments have begun to

seek methods of reducing the subsidies by introducing competition

among private sector operators. Where, as now in Lima or as in the

late 1980s in Santiago, this liberalization took the form of free entry

with little attempt at regulation of quality, many old and polluting

buses took to the streets. But where, as in many of the industrial

countries, the competition took the form of competition “for the mar-

ket”—that is, for the right of private companies to operate regulated

franchised services—air pollution was well contained. It is thus the

lack of quality regulation in privately operated bus sectors, not private

ownership per se, that is the cause of pollution.

The lesson is clear. Both unsubsidized public sector operations and

unregulated private operations can be very polluting. For poor coun-

tries that cannot afford heavy subsidies to public sector operations,

regulated competition for the market is an intermediate path that can

better reconcile affordability with service quality than either free entry

or public monopoly can.

F A Q 6


hicle, which can then operate more rapidly andsmoothly with shorter dwell times. When trunk linesare integrated, physically and in ticketing arrange-ments, with a system of feeder services they haveproved capable of maintaining or increasing the pub-lic transport share of trips even when incomes are in-creasing. The secret of their success has been that bothinvolved a combination of public infrastructure plan-ning with private operation that has made it profit-able for the private sector. This experience has shownthat good transport planning and service integration

is the essential prerequisite on which environmental

improvement of bus services has been founded.

One interesting characteristic of BRT systems isthat they have achieved their substantial environmen-

tal impacts without initial emphasis on advancedclean technology. The passengers are attracted by theaffordability, travel time, security, safety, comfort,cleanliness, and ease of use of the system. For ex-ample, conventional diesel bus technology has beenused in both Bogotá and Curitiba systems. In Bogotámore stringent environmental requirements on ve-hicles are proposed to be introduced in the next phaseof development. The lesson (IEA 2002) appears to bethat development of a BRT system has environmental

benefits by itself but can also provide an economi-

cally viable platform for the introduction of im-

proved technologies.

The Role of Mass TransitElectrically propelled transit modes are withoutdoubt the least locally polluting form of mass transit.That would include electric or fuel-cell powered roadvehicles as well as the more traditional electric rail-way. Very often, however, mass transit is thought ofonly in terms of rail-based systems. These have theadvantage of giving a perception of permanence andquality on the basis of which people in many large in-dustrial countries, particularly in Europe, have beenwilling to choose residential locations from whichthey can make their regular journeys to work by railrather than by car. Unfortunately these are very costly.Recent new underground railways have cost betweenUS$40 and $100 million per kilometer, which is be-yond the resources of most developing country cities(JJ&B Consultants 2001). Electric rail transport is very

clean at the point of use, but expensive.

The cost of rail transport is important to individu-als making these joint decisions on residential locationand choice of transport mode. In the absence of directcharges for road use (see International Experience 7 inchapter 5) many European cities have subsidized ur-ban rail transport heavily to compensate for its highcost. This has three main drawbacks. First, it is diffi-cult to get a sufficiently large road/rail fare chargingdifferential by increasing subsidies whereas in prin-ciple any size of differential can be obtained by rais-ing road charges. Second, whereas road pricing gener-ates revenue for the public authority, a public

Bus Priority Measures in London

Exhaust emissionProportion of reduction per

Measure buses affected bus affected

Peak period bus lane 5% 20%Contra-flow lane, all day 2% 35%Signal pre-emption 20% 12%Segregated busway 2% 60%Priority turns 5% 7%

Source: Bayliss 1989.

T A B L E3

Total segregation of bus lanes from the rest of the traffic, such as

what was done for TransMilenio in Bogotá, markedly reduces both

emissions and travel time.


transport subsidy is a drain on city budgets, which isparticularly important for typically cash-strapped cit-ies in developing countries. Third, a compensatingsubsidy to public transport has the effect of encourag-ing longer trip lengths, more sprawling developmentpatterns, and hence a higher long-term tendency torely on private transport. Rail mass transit subsidies

are likely to have only a weak influence on urban air

quality and should be relied on only in the context of

a comprehensive urban transport strategy.

Fortunately there are some more affordable masstransit alternatives. Where rail tracks exist, surface railor light rail transit systems can be constructed morecheaply, but usually have lower capacity (HalcrowFox and Traffic and Transport Consultants 2000). Seg-regated busway systems can yield almost equivalentmobility benefits at about 10 percent of the capital

cost for traffic volumes of up to 20,000 peak passen-gers per lane per hour in the peak direction, albeit at alower speed than rail mass transit (though in the rightcircumstances BRT may produce speeds that are com-parable with those of light rail transit). By combiningbusways with cleaner vehicle technology (such aselectric trolleys or buses fueled by CNG), large envi-ronmental benefits can be achieved without conflictwith system financial sustainability. For example, op-erators of the Bogotá TransMilenio system were re-quired to buy new, high-quality vehicles to competefor franchises, but claim that the increased efficiencyof movement has allowed them to vastly improve ser-vice and to increase their profitability without an in-crease in fares (Hidalgo 2001). More affordable mass

transit alternatives can often yield substantial envi-

ronmental benefits at much reduced cost.

65

Reducing Total Transport Demand

Shortening trip lengths, discouraging non-essentialtrips, and restraining private car use are some of themeasures that can be taken to reduce the overall de-mand for motorized transport. Use of private vehiclescan be limited both by pricing and by administrativeregulation.

Land Use Policy

The environmental importance of land use policyDemand for movement is derived from the demandto undertake activities at different locations. This de-mand can in principle be reduced by planning landuse in ways that minimize travel requirements, andparticularly minimizes private car movements, for agiven level of activity (Kenworthy and others 1999).The shape and geography of a city and the pattern ofdistribution of land use thus affect air quality (WorldBank 2002c). The establishment of a land use and

transport structure plan is the best basis for the ap-

plication of environmentally friendly transport devel-

opments.

Several different dimensions in land use planningcan be managed to influence urban air pollution:

� Density� Structure� Diversity� Local design.

DensityPopulation density affects motorized trips in twoways. First, for a given population, the higher thedensity, the shorter the distances between the loca-tions of different activities and the higher the numberof people who can comfortably walk to work, shops,or school. With high densities, multiple purposes mayalso be served by a single trip. Second, the higher the

density, the easier it is to provide frequent and easilyaccessible public transport services and thereby re-duce demand for private motorized transport. The“sprawling” of cities, particularly through artificialdiscontinuities in development resulting from compe-tition for activities between neighboring jurisdictions,tends to increase trip lengths and the need for privatemotorized transport. To control aggregate transport

emissions, it would be helpful to develop a policy

that increases, or at least maintains, the population

density.

In practice, the impact of increasing density on airquality has not always been positive because of inad-equate handling of traffic management. In South Asia,where urban densities have increased, the number oftwo- and three-wheelers has increased appreciably inpart because of their maneuverability in congestedtraffic. On account of the large number of operatorsinvolved, these vehicles are difficult to control foremissions as well as for traffic management. Poor

traffic management can negate to a considerable ex-

tent the potential benefits of good land use planning.

At first view, a densification strategy may appearto make larger demands on public funds than alower-density strategy. But this can be misleading. Inmany cases the additional investments required intrunk and distributional infrastructure to accommo-date an equivalent growth at low density in the pe-riphery may be even greater in aggregate, althoughthey may be hidden by the fact that they are distrib-uted among a number of agencies or budget heads.Developing country cities would be well advised to

attempt to estimate the infrastructure development

costs of alternative land use patterns.

StructureStructure is also important. It is easier to operate anefficient public transport system when the destination

C H A P T E R5


of the majority of trips is concentrated within the cen-tral business district (CBD). As a consequence, in cit-ies that are predominantly “monocentric” (most jobsand retail concentrated in the CBD), the share of tripsusing public transport tends to be higher than in“polycentric” (no dominant center) cities where theCBD contains only a small fraction of the total num-ber of jobs and retail shops. However, in concentrat-ing business and retail activities, it is important toavoid the creation of “hot spots” where pollutant con-centrations are extremely high. Policies that favor the

concentration of employment and retail activity in a

CBD can reduce transport-generated air pollution.

In reality, large cities are rarely purely monocen-tric, but have both a core CBD and a number of sub-centers with different activity compositions. In suchcities a large share of the trips to and from the CBDare likely to be by public transport, while those fromsuburb to suburb are more likely to be by privatetransport. This has generated the idea of concentrat-ing development around transit nodes, sometimes re-ferred to as transit-oriented development. As a broadguide, maintaining contiguous urbanization and ahigh-density CBD is not feasible for cities with morethan 5 million people. Therefore, a degree of

polycentricism should be allowed in megacities while

maintaining the primacy of the CBD to reduce trip

length and to maintain a high share of public trans-

port.

Moreover, as mentioned earlier, even if a citymoves toward the “right” structure, if traffic is poorlymanaged in the CBD, air quality may not improveand may even worsen. High densities andmonocentric city planning require a higher level ofprimary infrastructure investment and more rigorousmanagement and traffic law enforcement than are re-quired in lower-density, more dispersed cities. De-spite this, dispersing business outside of the CBD pri-

marily to save expenditures on traffic management is

not usually a cost-effective strategy.

DiversityPeople must live, work, shop, and be educated, usu-ally in different locations. Much traditional land useplanning was directed towards strict separation of

land uses, for environmental reasons, given the highlevel of air pollution caused by traditional heavy in-dustry. That separation was taken to its extreme in theplanning of development in many socialist cities ofEastern Europe where high-density residential devel-opment was put in peripheral areas of the city, andlinked to the main employment locations by high-vol-ume mass transit links. As industrial structure haschanged, industrial processes have become less pol-luting, and private transport has developed, the bal-ance between industrial- and transport-generated pol-lution has changed substantially. The more closelyactivities are co-located, the less will be the demandfor transport to satisfy them so that mixing land usesis therefore likely to minimize transport generated airpollution. In many cities the traditional separation of

land uses has become a net source of air pollution

rather than a protection against it.

Even now, the balance is not everywhere thesame. Coal-fired power stations and other heavy in-dustrial plants are the major sources of air pollutionin many Chinese cities. In contrast, some cities, suchas Curitiba, while locating their heavy industry awayfrom residential areas, have controlled total travel de-mand by mixing commerce, light industry, and resi-dential land uses and have influenced the choice ofmode by concentrating development in transit corri-dors. Judicious mixing of low pollution land uses can

be combined with the relative isolation of very high

polluters in a strategy to reduce urban air pollution.

Local designThe most polluting private motorized trips are theshort ones, due to high emission rates from coldstarts. Hence, disproportionate benefit can be ob-tained by designing cities to discourage such shortmotorized trips. Good safe provision for pedestriansand cyclists is particularly important as many shorttrips in middle-income developing country cities aretaken by car and taxi because of the danger and dis-comfort of walking or cycling. This should include theprovision and protection of pedestrian sidewalks, safewell-located crossing places on heavily traffickedroads, and adequate cycleways and cycle parking.Cities can reduce pollution from short motorized

CHAPTER 5. REDUCING TOTAL TRANSPORT DEMAND 67

trips substantially by good design of local facilities

for nonmotorized transport.

Problems of implementationDespite the importance of market forces in shapingcities in developing countries, planners can influenceurban structure both through regulation of privateland use as well as through public infrastructure in-vestments. Regulations can allow high densities, butthey cannot increase density if there is no demand.For that reason planners have sometimes attemptedto increase density by restricting land supply in orderto curb urban sprawl. The most common legal toolsused to curb sprawl are “green belts” and urbangrowth boundaries. But these often result in acutehousing shortages, particularly for the poor, and inhigh land prices, as in Seoul, Republic of Korea. Thelack of developable land may curtail the creation ofnew businesses and may have a negative effect on thecity’s economy. Land markets are often better than

land use regulation in determining the efficient level

of residential densities.

In a quite contrary direction, Indian cities such asBangalore have adopted regulations that attempt tokeep density low in the interest of avoiding conges-tion. These measures can also be counterproductive asthey typically lead to an imbalance between supplyand demand of land and push up average landprices. Regulations limiting density in the commer-

cial areas in the CBD should be critically reassessed.

The primacy of the CBD can often be maintainedby giving priority to the reinforcement of radial ser-vices (particularly public transport) over the construc-tion of multiple ring roads. A well-developed networkof public transport can also support the CBD andmaintain speed in the downtown area. While a goodorbital facility may be required to prevent truck trafficfrom crossing the CBD, excessive investment in arte-rial road capacity is likely merely to increase privatetransport trip lengths. Managing high densities re-

quires adequately structured supporting infrastruc-

ture investments, including those in expensive mass

transit in some circumstances.

Using land use planning instruments to reduce airpollution is challenging due to the divergent, and some-times incompatible, objectives of city governments.For example, urban planners may have to weigh thebenefits of creating employment opportunities in cit-ies by allowing new industrial plants to be set up withthe potential adverse effects of those industries andassociated traffic on air quality. It is crucial to coordi-

nate land use policies across key sectors, such as pub-

lic utilities, social services, urban development, and

transport, to realize air quality improvement.

Road PricingIn practice, the demand for movement is a functionnot only of the location of activities but also of thecosts of movement. Any form of subsidy of transport,whether public or private, will tend to increase triplengths by broadening the range of locations at whichit is affordable to satisfy particular demands. Correct-ing that by the introduction of road congestion pricingor some proxy, such as high fuel taxation or parkingcharges, may be essential to the creation of appropri-ate long-term incentives on activity location and tripdistribution decisions. Underpricing the use of trans-

port infrastructure is likely to have adverse effects on

urban air quality.

Where externalities of road traffic—both in theform of congestion and environmental impact—aresubstantial, the cost of road space to the road user islikely to be considerably below its marginal socialcost. While physical restraint measures have hithertoproven more acceptable than direct charges for roaduse both in industrial and developing countries, evenin industrial countries their effectiveness appears tohave been exhausted. Although only Singapore has athoroughly developed direct road charging system(see International Experience 7), the situation appearsto be changing as other cities such as London, Oslo,Stockholm, and Seoul have recently proposed to in-troduce or have introduced congestion charging sys-tems. For example, congestion charging, introducedin February 2003 in London, appears to have been


successful in shifting motorists to public transportand reducing journey times. Direct pricing of road use

has a high potential in developing countries both as a

means of generating local revenue and of reducing

congestion and air pollution.

Direct pricing (excluding fuel taxation) can in-clude charges for entering or traveling within a desig-

Congestion Pricing in Developing Countries

In September 1998, the government of Singapore replaced the manually enforced area licensing scheme (ALS)

that had operated since the mid-1970s with electronic road pricing (ERP). The new system generally covers the

same area as the ALS, but has been extended to approach and bypass roads. All vehicles are required to have an

electronic in-vehicle unit (IU) that accepts credit in the form of a smart card. Tolls are automatically paid when the

vehicle passes under a gate and a liquid crystal display indicates the current credit balance. Tolls do not fluctuate

in relation to actual traffic volumes but are adjusted quarterly to ensure optimum traffic speeds. The system cost

S$200 million (US$115 million at the exchange rate of September 1998) to implement, half of which was for the

free fitting of IUs. Once new traffic patterns stabilized, weekday traffic volume entering the restricted zone de-

clined 20–24 percent, from 271,000 vehicles per day to between 206,000 and 216,000. With those lower traffic

volumes, average traffic speeds in the zone increased from 30–35 km/h to 40–45 km/h. Improvements are less

clear on the three expressways in the ERP scheme, and the Land Transport Authority is reviewing ERP charges to

further optimize traffic flow.

Traffic congestion in Seoul, Republic of Korea, increased dramatically during the 1980s and early 1990s, despite

extensive construction of new urban freeway and subway lines. In 1996, the Seoul Metropolitan Government com-

menced charging 2,000 won (US$2.20) for the Namsan #1 and #3 Tunnels, two corridors with high private vehicle

use linking downtown Seoul to the southern part of the city. Private cars with three or more passengers and taxis,

as well as all buses, vans, and trucks, were exempted from charges, as was all traffic on Sundays and national

holidays. In the two years following commencement of the congestion pricing scheme for the Namsan #1 and #3

Tunnels, there was a 34 percent reduction in peak period passenger vehicle volumes. The average travel speed

increased by 50 percent from 20 to 30 km/h, and the number of toll-free category vehicles increased substantially

in both corridors. On the alternative routes, traffic volumes increased by up to 15 percent, but average speeds

also increased as a result of improved flows at signaled intersections linked to the Namsan corridors and in-

creased enforcement of on-street parking rules on the alternative routes. The whole of the annual revenue from

the two tunnels (equivalent to about $15 million) goes into a special account used exclusively for transport

projects, including transport systems management and transport demand management measures throughout the

city.

Other cities, such as Bangkok, Hong Kong, and Kuala Lumpur, have previously begun introducing road pricing but

have abandoned it in the face of political difficulties. As congestion increases, and other instruments to control it

fail, it is likely that they, among many others, will have to reconsider the pricing instrument that, as a by-product of

reducing congestion, reduces urban air pollution to an even greater extent.

Source: World Bank 2002f.



nated part of the city experiencing congestion (typi-cally the CBD), for use of selected road links, or forparking. Even Singapore—which has for many yearstaxed vehicle ownership very heavily and has been apioneer in charging motorists for traveling into thecity center—is now placing a greater emphasis on ve-hicle use than on restrictions on ownership. In the few

CHAPTER 5. REDUCING TOTAL TRANSPORT DEMAND 69

cases in OECD countries where direct cordon or areacongestion prices are charged, part or all of the rev-enues have been earmarked for public transport im-provements. For cities in developing countries that

lack resources to finance urban transport, the intro-

duction of direct charges might be expected to have a

double attractiveness: that is, as a source of finance

as well as an instrument of restraint.

Physical Restraint PoliciesPhysical restraints on vehicle use have been used bothin industrial and developing countries. The mostpopular restraint measures are schemes that limit useof vehicles on specific days according to their registra-tion plate number. Such schemes have been intro-duced in many cities, including Athens, Bogotá,Lagos, Manila, Mexico City, Santiago, São Paulo, andSeoul, both to reduce congestion and to improve theenvironment. They have achieved public acceptanceas a demonstration of commitment by government toreduce congestion and air pollution, and have provedless difficult to enforce than might have been ex-pected. In the short term they have achieved their ob-jectives (Bogotá reports 20 percent increase in averagetravel speeds). If well designed to discourage peakuse and coupled with public transport improvements,as in Bogotá, administrative restraint mechanisms

can at the very least give a “breathing space” to de-

velop more effective traffic restraint policies.

There are, however, obvious risks with this kind ofpolicy. It may encourage an increase in the number ofvehicles owned and induce more trips by permittedvehicles than would otherwise have been made. Inparticular, it may encourage the retention in operationof old, high-polluting vehicles that would otherwisehave been scrapped. Moreover, the experience ofBogotá suggests that careful design of the system—forexample, limiting the restriction to the peak time andhaving the proportion of non-use days sufficientlylarge to make it difficult to evade by the possession ofa single extra vehicle—may minimize the possibilityof counterproductive behavior. Physical restraint poli-

cies need to be accompanied by appropriate economic

measures to strengthen and maintain their effectiveness.

One aspect of restraint is particularly important. Itis politically difficult to restrict the movement of pri-vate vehicles unless there is obviously a viable alter-native. Both theory and practical experience indicatethat, at least in respect of trips to central areas, a co-

herent policy is likely to include a combination of car

restraint and public transport improvement.

Parking PoliciesParking policies have been widely and successfullyused in European countries, particularly to discour-age commuting trips by car. Such policies often com-bine limitations on the amount or location of parkingspace with high charges for long-term (commuter)parking. In developing countries this has often beenmore difficult to achieve because of the institutionalinability to control on-street parking. At the very least,in developing countries, avoiding the provision of

publicly funded free car parking can have beneficial

fiscal and environmental effects.

Parking policies have an impact on the effectivesupply of road space as well as on the demand for it.In many developing countries, highways and walk-ways are often encumbered with parked vehicles thatcongest traffic and as a consequence increase air pol-lution. Strong regulation to limit on-street parking to

locations where it has no effect on traffic flow is

likely to be the appropriate supply-side response in

most circumstances.

In order to reduce on-street parking, many citiesand countries impose minimum parking provision re-quirements in all new developments to create enoughoff-street parking space to cater to all vehicles wishingto access the development. The logic underlying thisis that as long as the costs of parking space are recov-ered through property rents, parking users can besaid to be paying directly or indirectly for the spaceallocated to parking. Unfortunately, road space andparking space are jointly demanded and the provisionof off-street parking space may actually attract newtraffic, thereby offsetting the gains from removingparked cars off the streets. If road space is providedbelow cost, then parking—a jointly demandedgood—should be charged more than its full costs in


order to avoid excessive vehicle use of roads. For thatreason, many industrial country cities use parkingpricing and availability as a demand-restraint mea-sure. The amount of parking in any area is then lim-ited to the maximum level considered necessary tosupport an “optimal” amount of road use. Pricingand parking supply regulation is used to implementthis strategy, which also implies specification of maxi-mum (rather than minimum) parking provisions fornew developments. Parking strategies for developing

country cities should be part of a comprehensive

transport plan and, particularly in central areas of

cities, should not necessarily aim at accommodating

all possible vehicle parking demands.

The Special Problem of MotorcyclesMotorcycles present some special problems, particu-larly in Asia. In many cities that have previously beenbicycle dominated, particularly in Vietnam and Tai-wan, China, motorcycle ownership and use havegrown at a phenomenal pace as incomes increase. Incongested cities, motorcycles are cheaper and quickerthan conventional public transport. As a consequence,the level of private motorized mobility in some rela-tively poor countries such as Vietnam is actually com-parable with that of many industrial countries.

At first sight increased motorcycle use might seemlike a very desirable outcome because motorcycles are

small, thereby occupying less road space, and muchmore fuel-efficient than other vehicle categories. Butaccident levels for motorcycles are very high and, incountries where two-stroke motorcycles are prevalent,they have a considerable negative impact on air qual-ity. In the short run, air quality impact can be reducedby programs to enforce the use of appropriateamounts of suitable lubricant and by relatively simpletraffic-management devices such as directional sepa-ration. In the longer term, the air quality impact ofmotorcycles can be reduced most substantially by ashift to four-stroke motorcycles, as in India, Vietnam,and Taiwan, China, or to direct injection two-strokeengines. Stringent technical regulation is necessary to

control the environmental and safety impacts of traf-

fic in motorcycle-dominated cities.

The real problem arises in the longer term. Withfurther increases in income, and in the absence of ad-equate public transport alternatives, motorcycles arelikely to be progressively replaced by private cars. Inmany of the Asian cities with relatively low propor-tions of space devoted to roads, this will generateheavy congestion and have an even greater negativeimpact on air quality. Despite the great political diffi-culty in restraining freedom of movement of motor-cycles, it will be essential to introduce restraints on

private motor vehicle movement at an early stage if

city environments are to be protected against the al-

most inevitable shift from motorcycles to cars.

71

Designing a Supportive Fiscal Framework

The fragmentation of decisionmaking in transportamong so many private and financially motivatedagents (including private individuals) puts great em-phasis on getting the prices right. Where there are un-charged-for external effects, as in the case of air pollu-tion, it is likely to be necessary to make appropriateadjustment of relative prices through fiscal measures.This may take the form of taxes, subsidies, or directcapital expenditures. An appropriate set of support-ing fiscal policies can increase the effectiveness of ad-ministrative and regulatory measures. Developing

country cities may improve the effectiveness of their

attempts to reduce air pollution by using existing pow-

ers or by seeking new powers to use fiscal instruments.

Direct Taxation on EmissionsThe economic ideal would be a system of direct taxa-tion on emissions. If such a system could be devised,it would not be necessary to have complex adminis-trative controls, because there would be strong incen-tives for all agents to limit their emissions levels.Choice of vehicle and fuel technology would both bedriven straightforwardly by these economic incen-tives. Insofar as the level of emissions is determinedby the vehicle design (including the fuel for which itis designed), taxation on the capital cost of vehicles isan approach along these lines. For vehicle-manufac-turing countries it would also be possible in principleto use a system of tradable emissions certificates toharness market incentives to encourage economic,clean technology. For non-manufacturing countries itcould be embodied in import duties that vary accord-ing to the vehicle’s emission characteristics. However,even this does not affect the way in which vehicles aremaintained and operated, and it would still be neces-sary to have some other complementary taxation onthe vehicles in use.

Fuel TaxationUnfortunately, for vehicles in use, the continuousmeasurement of emission levels and application ofvariable taxation levels are technically a long way off. Itwill therefore be necessary to devise the best proxy tax ortaxes. Because many pollutants are emitted in rough pro-portion to the amount of fuel burned, fuel taxation is anobvious candidate (Gwilliam and others 2001).

The incentive properties of fuel taxationOne way of assessing the potential of transport fueltaxation as an environmental instrument is to look atits efficacy with respect to reducing vehicle km trav-eled, fuel consumption per km, and emissions perunit of fuel consumed:

� Reducing overall vehicle km traveled. Hightaxation on transport fuels will encourage a re-duction in trip numbers and trip lengths as wellas favor public over private transport modes.Although the mode shift effects themselves maybe small because of the range of different di-mensions of adjustment, the overall price elas-ticity of demand for gasoline may be between–0.5 and –0.8, although that for diesel is likely tobe considerably lower.

� Reducing fuel consumption per km. A high taxon fuel will encourage the use of more fuel-effi-cient vehicles. But the level of congestion is themost important factor in determining fuel con-sumption, and the fuel tax is not very efficient asa charge for congestion.

� Reducing emissions per unit of fuel con-

sumed. The high degree of differentiation of en-vironmental damages from the same fuelsacross various users, technologies, and locationslimits the effectiveness of fuel taxes for control-ling air pollution (Lvovsky and Hughes 1999).

C H A P T E R6


Fuel tax has good incentive properties for re-

ducing the amount of vehicle kilometers traveled

and encouraging the use of fuel-efficient vehicles,

but it fails to reflect the location of emissions.

The multiple objectives of fuel taxationIn many developing countries, fuel taxation is usu-ally the responsibility of central government andtaxes on hydrocarbons can account for as much asone-fifth of all central government tax revenue(Bacon 2001). These taxes are usually considered areliable revenue source because fuel has a lowoverall elasticity of demand with respect to priceand the taxes can be collected cheaply. But theyusually are expected to fulfill at least four functions:

� Revenue function. They are primarily aimedat raising revenue for general (non-trans-port) expenditure purposes.

� Road user charge function. Taxes on trans-port fuels are often the primary meansthrough which vehicles can be charged forthe use of roads, and some part of fuel taxrevenue is often earmarked for financingroad provision and maintenance.

� Redistributive function. As part of centralgovernment policy, their redistributive char-acteristics might also be of great importance.

� Environmental function. In the absence ofdirect charges for air pollution they may beexpected to be a surrogate environmentalcharge.

It is clearly not possible to achieve so many ob-jectives simultaneously and efficiently through asingle tax. The challenge of satisfying multiple ob-jectives is especially difficult in low-income coun-tries, where fewer policy instruments are avail-able. In the use of fuel taxes, compromises have to

be made between the effects on government rev-

enue generation, income distribution, the efficient

use of roads, and air pollution.

The principles of fuel taxationSome useful guidance for evaluating these com-promises can be obtained from the general prin-

ciples of optimal commodity tax theory, which focuson minimizing the loss of welfare to consumers inraising a given sum of money for the governmentthrough commodity taxation (Newbery and Stern1987). A fundamental principle is that in order to

avoid economic distortions taxes should, as a rule, be

levied on final consumption goods rather than on in-

termediate goods. This principle suggests that fueltaxation should be concentrated on gasoline, which isused predominantly by private cars as a consumptiongood, rather than on diesel, which is used in largequantities by freight and public transport vehicles as aproducer good.

Such a prescription, however, has three major dif-ficulties. First, not all consumption goods (for whichdiesel is an input) are taxed and it may be necessaryto tax the inputs instead. Second, freight transport ofproducer goods generates air pollution, but operatorshave less incentive to use less fuel (lower fuel con-sumption would normally lower overall emissions)when fuel is cheap. However, by far the greatest prob-lem with differential fuel taxation concerns the effectsof inter-fuel substitution. In the long run, diesel, gaso-line, CNG, and automotive LPG are all technologi-cally possible substitutes. The common combinationof a high gasoline tax and a low diesel tax may en-courage vehicle owners to switch from gasoline todiesel when they buy or replace light-duty vehicles(as has already happened in some countries such asFrance, and is being advocated for fuel economy rea-sons in the United States).1,2 While “clean diesel” inthe EU and North America may mitigate the impactof such fuel switching, the same phenomenon in de-veloping countries would most certainly mean muchhigher particulate emissions.

1A recent study by M. Cubed, commissioned to provideeconomic and technical input into California Assembly Bill2076, argues that because diesel passenger vehicles are typi-cally 35 to 50 percent more fuel efficient than similar-sizedgasoline vehicles, a shift to diesel vehicles could substan-tially reduce the state’s dependence on petroleum. The fullreport is available online at www.dieselforum.org.

2To avoid this anomaly, a low tax on diesel fuel mightbe supplemented by a high tax on light-duty diesel vehicles,particularly those used primarily in intra-city transport.

CHAPTER 6. DESIGNING A SUPPORTIVE FISCAL FRAMEWORK 73

A complementary principle is the well-known“Ramsey pricing” rule, which proposes that tax rateson consumer goods be so set as to be inversely pro-portional to the goods’ own price elasticities of de-mand in order to minimize the overall loss of welfare.Normally, application of this principle in its pureform is likely to be distributionally perverse, since thedemand for most basic necessities (such as staplefoods) is inelastic, while that for nonessential goods islikely to be more elastic. However, this is less of aproblem in urban transport. Demand for fuel for pri-vate cars is believed to be (mildly) inelastic. Given theconcentration of car ownership and use in the upper-income groups in developing countries and weak sys-tems for direct taxation, a high incidence of taxationon gasoline makes for a very progressive tax. Hence,if the basic distribution of income is viewed as inequi-table, indirect taxes may be structured to have agreater impact on the goods that make up a relativelylarger share of the budgets of higher-income house-holds than on the goods that are more important forlow-income households. For these reasons, relatively

high taxes on gasoline can be both economically effi-

cient and distributionally attractive.

Not all efficiency arguments militate in the direc-tion of high differentiation between gasoline and die-sel tax rates. In principle, users ought to pay the long-run marginal costs of road use, including the costs ofcapital. Many developing countries have poor-qualityroad systems because of underfunding of mainte-nance. A fuel tax, or surcharges on the fuel tax that aredesignated specifically as road user charges, may bethe most obvious and acceptable proxy for directcharging, especially when the revenues are trans-ferred directly to a user-managed road fund(Gwilliam and Shalizi 1999). As wear and tear iscaused primarily by heavy vehicles, which are fueledprimarily by diesel, this suggests that diesel tax mightbe an appropriate proxy for direct road maintenancecharges. Diesel tax, however, has serious shortcom-ings in this respect. It does not accurately reflect theroad deterioration caused by different vehicle catego-ries and provides inefficient signals on vehicle sizeand weight. Even within the automotive diesel fleet,

a tax on diesel needs to be supplemented by some

charge on vehicle axle loadings, preferably levied on

the basis of distance traveled.

From the point of view of air quality, diesel ve-hicles are typically much more damaging than gaso-line vehicles, especially when compared to gasolinevehicles equipped with three-way catalytic convert-ers, an increasingly common phenomenon as devel-oping countries move to 100 percent unleaded gaso-line. Heavy-duty vehicles worldwide run on diesel,but light-duty vehicles can run on either gasoline ordiesel. For vehicles of comparable size, diesel vehiclesare more expensive to purchase, but if diesel tax ismarkedly lower, fuel cost savings can compensate forthe higher purchase price of diesel vehicles. Diesel ve-hicles are especially suitable for high annual km ve-hicles. Diesel fuel should be properly taxed in order toreflect the marginal social damage caused by increas-ing air pollution. This is likely to mean that in nearly

all developing countries, the tax on diesel would need

to be raised to capture marginal social damage per li-

ter of fuel.

Aside from political opposition from truckers andother heavy users of diesel, one concern about in-creasing diesel tax is the impact on the economy, andon the poor in particular. The share of expenditure onall fuels as a percentage of total household expendi-ture consists of direct consumption of fuels and indi-rect consumption through purchases of goods andservices that have fuels as inputs. Direct consumptiontends to be concentrated at the top of the income dis-tribution for both gasoline and diesel, but the indirecteffect is more important for lower income groups fordiesel because it is an important input to many goodsand services. Studies that have examined the impactof raising diesel tax on household expenditures in de-veloping countries have shown a modest impactwhich is mildly regressive—that is, the total expendi-tures of poor households rise more in percentageterms than those of the rich when the price of diesel isincreased, although these effects are small. Because of

the significant impact of higher taxation on non-au-

tomotive uses of diesel—in rail transport, agriculture,

and industry, for example—it may be sensible to give

rebates on the higher diesel tax to industrial and agri-

cultural users of diesel.


Taxation on VehiclesGiven the complexity just outlined, it is not possibleto structure taxation on fuels to simultaneously satisfyall of the policy objectives. For instance, fuel taxationdiscriminates relatively poorly between vehicles ofdifferent axle weights, which is critical in determiningthe amount of road damage done by vehicles andhence the appropriate charge for their use of roads.For that reason differentiation of annual license dutiesamong vehicle categories is common. Where that isthe case, it is sensible to consider the possibility of

using differentiation of vehicle license duties to en-

courage cleaner technologies.

Another common source of revenue for govern-ments is import duties on vehicles. Usually the levelof such duties is motivated either by revenue maximi-zation or by protection of local manufacturers. In ei-ther event, the result is typically that duties are basedon the market values of vehicles so that those on newvehicles are greater than those on old, and duties onsophisticated vehicles greater than on those of simplertechnology. Even if duties are differentiated to reflectdiffering exhaust emission levels, different age limitscan provide a loophole. In Sri Lanka, for example,diesel vehicles carry higher import duties than gaso-line vehicles, but the age limit is three for cars andfive for dual-purpose vans, so that there has been asteady shift from gasoline cars to diesel vans (Cam-bridge Economic Policy Associates 2002). The overalleffect is to discourage best-practice technology froman air quality viewpoint. The impacts on air quality

should be borne in mind when determining the struc-

ture of duties on vehicle imports.

Similar considerations apply to physical measureson imported vehicles where it is common to be morerestrictive on the import of new vehicles than old. Insome countries importing secondhand engines andchassis and putting new bodies on them is one way ofbypassing emission standards for new vehicles. While

maintaining type approval control of import of new

vehicles, it is important to concentrate effort on con-

trolling imports of old vehicles or rehabilitated com-

ponents.

Constructing a Road TransportTax PackageIn the absence of other direct charges for road use,pump prices of transport fuel should cover the re-source cost of the fuel, the costs of road use (both roaddamage and occupation of road space), and the envi-ronmental costs associated with the fuel use if not oth-erwise charged for (some costs might be recoupedthrough differential vehicle taxation). Although fueltaxes can strongly affect fuel consumption patterns,they have other significant welfare impacts, includingspillover effects. Some basic guidelines for fuel taxingfollow from this:

� In addition to fuel taxes, more precisely targetedalternatives (such as differentiated vehicle taxesand road user charges) should be consideredwherever possible, in view of the limitations offuel taxes in achieving multiple objectives.

� Environmental externalities should be correctedfor by taxing polluting vehicles and fuels, not bysubsidizing less-polluting alternatives. If the ob-jective is to achieve a certain amount of pollu-tion reduction by reducing demand for pollut-ing goods, subsidizing less-pollutingalternatives costs society more than increasingtaxes on polluting goods.

Sale of imported secondhand diesel engines is common in develop-

ing country cities.

CHAPTER 6. DESIGNING A SUPPORTIVE FISCAL FRAMEWORK 75

� There is a strong case for setting the gasoline taxabove the general tax rate on commodities. Richhouseholds spend a higher proportion of theirbudgets on gasoline than do poor households,gasoline vehicles give rise to a number of exter-nalities, and emissions from gasoline enginesmay affect poor households more than richhouseholds.

� There is also a strong case for a diesel tax. Al-though some diesel is used as an intermediategood, the taxation of even this segment of themarket is justified if it is the principal way ofcharging heavy vehicles for wear and tear onthe road and if the final goods (for which dieselis an input) are not necessarily taxed.

� Given diesel’s high long-run substitutability forgasoline in light-duty vehicles and its stronglynegative externalities in urban areas, tax policiesfor petroleum fuels should take careful accountof, and minimize, the possibilities for sociallyundesirable interfuel substitution, such as theso-called dieselization of light-duty vehicles.

Property Taxation and FeesThe land use objective to minimize transport emis-sions is to promote contiguous and dense develop-ment. A standard ad valorem property tax is probablythe best fiscal tool to achieve this. Confiscatory capitalgains taxes on real estate, while appearing equitablebecause they concern “unearned income,” in fact have adisastrous effect on land use efficiency. Increasing capi-tal gains tax increases the threshold for the profitabil-ity of land conversion; as a result, obsolete, inefficient,and low-intensity land uses are maintained for a muchlonger time than they would be in the absence of suchtax. Impact fees on new construction, whether forbusiness or residential purposes, should cover thepublic costs of improving infrastructure (roads, drain-age, sewerage, public utilities). Insofar as these arehigher for dispersed and greenfield site developmentthan for densification or infill, such fees would dis-courage urban sprawl. Tax incentives may be impor-

tant in influencing the land use structures that, in

their turn, influence transport demand and emissions.

Public Expenditure Policies

Environmentally oriented investment policiesFiscal policies include policies on public expenditureas well as on taxation. Taxes, road tolls, parkingcharges, tax rebates on public transport fuels, andcapital investment in transit infrastructure may all bedirected at achieving environmental ends. The mostimportant consideration is that any such allocationneeds to be justified in terms of the benefits that it ac-tually achieves. Air pollution reduction is one of the

benefits that should be systematically taken into ac-

count in assessing urban public expenditure priorities.

Public transport subsidiesThe imposition of stringent emission and other ve-hicle standards, without simultaneously introducingbus priority measures to improve efficiency, tends toincrease capital costs without offering compensatingreduction in operating costs. This raises the problemof the financial sustainability of service. Many indus-trial country cities subsidize public transport fares.While this may seem to be a reasonable approach, it isnot generally cost-effective as an environmentalpolicy. First, there is strong evidence that up to half ofany subsidy “leaks” to benefit transport industry in-terests including owners and employees. Second, be-cause most car owners’ use of their vehicles is not sen-sitive to public transport fare levels—cross-elasticitybeing of the order of 0.1—there is little shift of travel-ers from private to public transport. Third, becausethe modal shift is small but the subsidy is paid to all,subsidizing public transport fares, however desirable

on other grounds, is not likely to be a cost-effective

way to reduce environmental impact.

Given the poor public image of public transport,subsidies to improve public transport quality mightbe somewhat more effective in affecting mode choice,particularly in countries where car ownership is re-stricted to the relatively rich. But even for higher-in-come groups, regulatory reform to encourage expressor air-conditioned buses to attract car owners is likelyto be more effective than giving subsidies, as experi-ence has shown in cities of varying average incomelevels such as Bangkok, Buenos Aires, Dhaka, and


Seoul. Some liberalization of market entry is likely to

be the most effective way of developing public trans-

port services to attract patronage from private cars.

Subsidies on clean fuelsThe most direct approach would be to subsidizecleaner vehicles and fuels. However, the need to en-sure financial sustainability applies. The cleaner ve-hicles must be capable of being operated reliably andeconomically. This may require not only an initialcapital subsidy but also substantial investment intraining and maintenance facilities for the new tech-nology, as well as fiscal effort to keep the price of thefuel attractive. If that is the case, then it is necessary

to ask which alternative policies (for example, invest-

ment in busways) might have been introduced at the

same cost to the government as the fuel-duty loss in-

volved in subsidizing clean fuels.

Several countries follow the policy of keeping fueltaxes relatively low and giving positive incentives forthe use of relatively cleaner fuels. This has been theapproach of promoting fuel from agricultural crops inthe United States and promoting CNG in a number ofcountries. But such measures can be expensive to thepublic budget and, to the extent that fuel prices thatconsumers face are lower, encourages excessive use offuel. As far as externalities are concerned, the gener-ally accepted philosophy (Sandmo 1975) is thereforethat tax rates on goods that have external costs

should be adjusted upward to reduce their consump-

tion to a social optimum, and any additional revenue

collected used to adjust general tax rates downward.

77

The Supporting Institutional Framework

The Range of Institutions Involved inUrban Air QualitySteps taken to tackle mobile sources of urban air pol-lution affect, and are affected by, a wide range of sec-tors and institutions. Transport, petroleum, environ-ment, finance, urban planning, and even agriculture(for biofuels) are among the sectors involved. A num-ber of groups are engaged to varying degrees inpolicy formulation and implementation, including theauto and oil industry; goods delivery and passengerservice firms; vehicle repair and service industries;manufacturers of emissions measurement equipment;environmental nongovernmental organizations(NGOs); firms with their own fleets; private vehicleowners; municipal, state, and national governments;and regional trade organizations. Air quality policy,more than most other policy areas, needs a high levelof integration with, and within, the policy of agenciesfor which air quality is not the central issue of con-cern.

The Role of Central GovernmentCentral government has a number of functions criticalto the establishment and implementation of an airquality improvement strategy at the city level.

First, central government is usually the onlyagency that is commissioned to set policy and stan-dards for fuel and vehicles. Municipal governmentsmay be able to impose tighter standards in somecases, but they cannot “relax” the standards set bycentral government even if that would mean muchmore effective enforcement in practice. Vehicle inspec-tion is a good example. A number of countries haveregulations that require that every vehicle in the coun-try be inspected regularly. In reality, only a small frac-tion of vehicles are inspected, using tests that can be

easily cheated on. In order to make inspection moreeffective, an obvious approach is to employ more ef-fective test procedures and quality-control measuresand target vehicle categories likely to contain a dis-proportionately large fraction of high-circulationgross polluters in large cities. But such a selective ap-proach with test protocols different from those set bycentral government would normally not be permit-ted. When delegating the enforcement of environmen-

tal standards to municipal agencies, national govern-

ments should carefully consider likely probability of

effective enforcement in light of financial and human

resources available.While government should consult industry and

civil society about vehicle emission and fuel stan-dards, it should not outsource responsibility for policydevelopment and standard setting. If refinery re-vamps are involved, a lead time of several years istypically necessary. New investment may be difficultto attract if there is a great deal of uncertainty abouthow rapidly tighter standards would be adopted inthe future or if government decisions are reversedfrom time to time. What is especially important in

this respect is the establishment of a predictable and

consistent policy and regulatory framework that will

help attract the private sector financing needed to

implement fuel quality and vehicle technology im-

provement.Other regulatory frameworks may be equally im-

portant. For example, the ability to increase publictransport operating efficiency through competitivetendering of franchises, including the imposition ofstrict environmental standards on contracting suppli-ers, depends on the existence of an appropriate legalframework for franchising. Further, franchising ar-rangements can be designed to positively encouragerelatively non-polluting operators. Framework legis-

C H A P T E R7


lation on urban public transport should pay explicit

regard to its air pollution impacts.Central government also usually has the ultimate

power in fiscal matters. The level and structure of fueltaxation, which is usually a central government pre-rogative, is most important as an inducement to envi-ronmentally sensitive decisionmaking on the choice oftechnology as well as on the amount of transport de-manded. An example is that maintaining a high taxdifferential between gasoline and diesel to raise rev-enue from the relatively well-off (namely, gasoline carowners) may simply stimulate the purchase of morepolluting diesel vehicles in developing countries. It is

important that taxation powers be exercised with en-

vironmental, as well as revenue generation, consider-

ations in mind.There may also be a range of direct expenditures

that, because of their public good characteristics, arelikely to be undertaken only if centrally financed.These may include expenditures on environmental re-search, central laboratory facilities, and environmen-tal education and information programs. Central gov-

ernment should accept the responsibility for

encouraging clean air by having a policy on clean air

related public expenditures.Concerted action within and among actors in dif-

ferent sectors is important for optimal policy formula-tion. Transport, finance, environment, oil and gas, andurban planning are principal sectors that have a stakein, as well as a direct influence on, mitigation mea-sures to control transport-related air pollution. Insome cases the problem is simply a matter of lackof coordination; in others there may be clear con-flicts of interest between different ministries oragencies. At the national level, for example, the en-vironment ministry may want a low tax on CNG,whereas the finance ministry may not want transportfuels to carry a low tax, especially if a low tax onCNG means large-scale switching from high-tax gaso-line to CNG. It is important to encourage close col-

laboration among the national ministries of environ-

ment, transport, finance, and energy on air pollution

strategy.

The Hierarchy of Government andInter-Jurisdictional CollaborationThe division of function between levels of govern-ment on a general subsidiarity principle (that powersshould be devolved to the level in the hierarchy ofgovernment at which they can be most effectivelyimplemented) is generally sensible. Sensitivity to thatfact may require the granting of some powers to localauthorities to levy surcharges as part of local environ-mental policy or to be able to retain as local tradingincome the revenues of local road pricing schemes.Where responsibility for urban transport is decentral-

ized, appropriate fiscal arrangements must be made

to facilitate local ability to meet those responsibili-

ties.

But fragmentation of responsibility between hier-archies of government may also cause problems. Ifdifferent cities with comparable air pollution andpopulation exposure allocate varying amounts offunds for air quality monitoring and mitigation mea-sures, then some large cities that have chosen not toallocate much funding for air quality monitoringcould end up not collecting even basic data on ambi-ent air quality. In the absence of such fundamentaland critical data, it would be difficult for central gov-ernment to formulate a sensible air quality manage-ment strategy. Another example can be found wheredifferent municipal governments pursue their ownversion of vehicle I/M. As annex 7 argues, there areconsiderable merits to centralized software develop-ment and standardizing test protocols and equipmentspecifications. Not having the same air quality moni-toring equipment specifications and procedures is yetanother example of collected data not being condu-cive to analysis and inter-city comparison. Standard-

izing equipment specifications and data collection

procedures across major cities in the country, and en-

suring that essential data be collected in very large

cities, would be useful for policy formulation at both

the national and municipal level.

In public transport, the division of responsibilitybetween the state and municipalities for the provision

CHAPTER 7. THE SUPPORTING INSTITUTIONAL FRAMEWORK 79

of bus services has inhibited coordination of servicesin many Brazilian conurbations. Similarly, competi-tion between independent municipal jurisdictions incontinuous conurbations has often resulted in “beg-gar-my-neighbor” policies that are seen to be in theshort-term advantage of the individual jurisdictionbut that, when pursued by all, make all worse off.1

Unwillingness to pursue restrictive parking policiesor, even worse, the provision of subsidized parking toattract trade, is a prime example.

There are several different institutional ways ofconfronting this. In a number of countries in the in-dustrial and developing worlds, there are single-pur-pose, conurbation-wide agencies for urban transportplanning and management that have priority in stra-tegic decisions over the separate jurisdictions in urbantransport matters (World Bank 2002f). In others, thereare formal consultation and collaboration arrange-ments.2 It is important to make the environmental ef-

fects of urban transport one of the responsibilities of

urban transport and land use institutions even where

there is a parallel institutional responsibility for air

quality protection.

The Organization of MunicipalGovernmentAir pollution problems are location-specific. Oncestandards are set, particularly if there is geographicdifferentiation, it is state and municipal governmentsthat act to implement them. Governments at theselevels monitor air quality, emission and discharge lev-els, and fuel and lubricant quality; integrate air qual-ity considerations into overall city developmentplans; develop traffic flow, demand management, andother strategies for mitigating traffic congestion andemissions; and, where appropriate and fiscally pos-sible, offer financial and other incentives to facilitatecleaner technology and fuels, retirement of old equip-

ment and vehicles, and other means of mitigating airpollution.

Problems arising from different objectives in dif-ferent ministries and departments need to be ad-dressed. At the national level such conflicts may bewell aired and find their resolution in high-level cabi-net decisions. At the municipal level, the police, cityplanning, transport, and other agencies are frequentlyless well coordinated. For example, the police maywish to give priority to cars at junctions to keep trafficmoving while the transport department may wish togive priority to buses to attract passengers to publictransport. This type of conflict is accentuated wherethe police report directly to a central ministry ratherthan through the local authority structure. The pau-city of adequate technical advice and the absence ofan effective political representation of some sector in-terests in mayoral decisions may also result in subop-timal resolution of such conflicts.

The institutional requirements for effective trafficmanagementTraffic management measures are relatively cheapand quick-acting, but they are not a guaranteed, one-shot cure for traffic congestion. They need effectiveplanning, implementation, and enforcement skills anda high and continuing degree of political, institu-tional, and human resource commitment, all of whichtend to be in short supply in developing countries.Fundamental to the successful implementation oftraffic management measures is the establishment of atraffic management unit at the local government levelwith the consolidated authority and ability to planand implement suitable traffic management schemes.The role of the police in complementary enforcementactivities is also essential. The traffic management sys-tems implemented in Manila and Mumbai in the1980s are now largely out of commission because ofinstitutional failures. Institutional factors are critical

to the success of traffic management.

Conflicts of interest frequently arise over trafficmanagement. At the technical level, traffic signal set-tings for delay minimization differ from those foremission minimization. At the political level, conflicts

1This is a clear case of the “prisoner’s dilemma” phenomenonas expounded by Luce and Raiffa.

2Such as the coordination committees set up in the major Bra-zilian conurbations to bring together state and municipal inter-ests.


of interest may occur among jurisdictions competingfor business because restraining parking freedom candiscourage trade in their area. In fact, available evi-dence in western Europe suggests that city centerpedestrianization actually increases trade in the con-trolled area, and can be facilitated by regional collabo-ration to prevent misplaced “beggar-my-neighbor”policies to attract trade (that is, hoping that other dis-tricts, but not one’s own, will restrain parking). Con-

flicts of interest can be minimized by better informa-

tion and institutional coordination.

At the municipal or city region level the solutionmay be found in the establishment of a department oragency with comprehensive responsibility for estab-lishment of a transport and land use plan, with airquality as one of its objectives, to which contributorytechnical functions (traffic management, public trans-port policy, road investment) must conform. Ex-amples of such agencies are to be found in most of theUnited States and European cities that have well-inte-grated urban transport policies. It is important to es-

tablish a specific responsibility for urban air quality

within the municipal government structure.

Involving the Private SectorMany of the agents that implement activities affectingurban air quality are in the private sector. They in-clude fuel and vehicle manufacturers and marketers,public transport companies, industrial and commer-cial firms operating their own transport fleets, and, ofcourse, individuals using private transport vehicles.For an air quality policy to be effective, all of these

agents must have some incentive to behave in a way

consistent with the policy.Those incentives may be financial, where taxes are

used to make socially desirable behavior individuallyattractive; legal, where penalties are attached to non-observance of environmental standards; or moral,where antisocial behavior is sufficiently well identi-fied and advertised for peer pressure to influence be-havior. In all of these dimensions, but particularly inthe last, the incentives are likely to be more powerful

if the agents themselves have been involved in the

formulation of the policies.

The private sector can play a positive role, pro-vided that certain conditions are met. If rules andregulations appear to be established without ad-equate consultation or at short notice, it might becomedifficult even for efficient private sector firms to com-ply and stay in business. If firms are not treatedequally—for example, if subsidies are given only tostate-owned enterprises but not to private sectorfirms—it would be difficult for private firms to com-pete efficiently. If standards and regulations areweakly enforced, there could be partial or total degra-dation of the market whereby those who “cheat” endup driving out those who do not. For the private sec-

tor to play a meaningful role in air quality improve-

ment, the government needs to create an enabling

policy environment whereby there is a clear and

transparent regulatory framework, a level playing

field, and enforcement of standards and regulations

on all actors.

Vehicle manufacturersVehicle manufacturers can be involved in discussionsabout technically feasible emission or fuel economystandards, such as the CAFE standards in the UnitedStates or the more recent European voluntary fuel-economy agreements. The auto fuel policy report inIndia, completed in August 2002, involved extensiveconsultation with government agencies, vehiclemanufacturers, refiners, and research institutions toset future fuel quality and vehicle emission standards,thereby greatly enhancing the probability of compli-ance at the manufacturing level.

The possibilities of perverse adaptation to theregulations need to be carefully protected against.Both manufacturers and consumers may be expectedto find ways around standards that they find restric-tive. For example, the exclusion of SUVs and vansfrom the light-duty vehicle emission and CAFE stan-dards has permitted their rapid growth in the UnitedStates, significantly reducing the effect of the stan-dards on both local and global emissions. This hasbeen exacerbated by the considerable tax exemptionenjoyed by this vehicle category. It is very important

to look out for loopholes in the specification of stan-

dards.


In some developing countries, there are significantenvironmental risks attached to the protection of a do-mestic manufacturer embodying seriously outdatedtechnology, as in the case of the Ambassador car in In-dia until the opening of the market to new suppliers.Other ways of involving the manufacturers, such asrequiring them to guarantee the durability of perfor-mance of their vehicles to the agreed standards,would be extremely useful. Durability requirementsare rare in developing countries and require signifi-cant strengthening of capacity to enforce in-use ve-hicle emission standards by means of regular vehiclemaintenance and fuel quality standards at the point ofsale, but government and the private sector should

work together to adopt durability requirements that

are universally implemented in industrial countries

today, and considered essential for controlling ex-

haust emissions.

The implementation and enforcement of stringentstandards may be an effective way of “forcing” tech-nological development in countries that manufacturevehicles or refine fuels. In both India and Taiwan,China, the governments have progressively tightenedemission controls on motorcycles, with the effect thatfour-stroke engines have replaced two-stroke in newvehicle manufacture in Taiwan, China. At the sametime, increasingly tighter standards have also led toimprovements in the two-stroke engine technology it-self as well as the development of suitable catalyticconverters. Setting stringent standards when there are

cost-effective, commercially proven technology alter-

natives to meet the standards can be instrumental in

forcing out old polluting technologies.

The refining industryParticularly in the case of sulfur reduction, tightenedstandards impose high capital costs on the manufac-turers. If the industry considers the time-frame set bythe government or the level of emissions reductionstargeted to be unrealistic, policy implementation islikely to encounter delays, or worse, result in loss ofcredibility for environmental regulations. In onecountry, a number of tighter fuel quality specificationshave been issued with specific implementation dates,only to have the dates missed time and again and re-

vised dates issued because the government-owneddomestic refineries were in no position to producetighter specification fuels. Close consultation with the

refining industry is an important component of effec-

tive fuel policy in countries with domestic refineries.

There are usually different ways of meeting thesame emission standards. The diesel certification pro-cess in California allows refiners to come up withmuch cheaper alternative diesel formulations to the“reference” diesel (defined by the state government),as long as they meet the same emission levels. Thisapproach ultimately saves consumers money (see box3 in annex 2). While the particular certification pro-cess is complex, the basic principle nevertheless offersa useful lesson: whenever possible, suppliers should

be engaged in the standards discussions and be given

the flexibility to seek the lowest-cost options for

meeting specified emission targets.

One sector distortion that has an adverse environ-mental impact is protection of government-ownedmonopoly oil companies, which tend to acquire con-siderable political power, especially in oil-producingcountries. This has resulted in major oil producerscontinuing to supply leaded gasoline on the domesticmarket when countries that are much poorer havelong since banned it. Addressing this sector distortionrequires a strong political will, and often occurs onlyafter there is a major change in government, as in Ni-geria in recent years. Given the damaging economy-

wide impact of protection of state-owned monopoly

oil companies in many countries, concerted efforts

should be made to alleviate this sector distortion.

Public transport operatorsPublic transport companies are often operating thelargest fleets of heavy and high-annual-km vehicles.In many countries these vehicles are manifestly ill-maintained and likely to be highly polluting. They areeasy to identify, and can be subject to environmentalcontrol, even where there are not adequate facilities toregularly test all vehicles. But it must be rememberedthat the main objective of public transport policy issustainable provision of mobility, usually to lower-in-come segments of the population. The improvementof environmental quality is an important, but not pri-


mary, consideration. Hence it is important that mea-sures to improve the environmental quality of busesbe part of a comprehensive public transport regula-tory regime. Evidence in cities such as Lima, wherethere is relatively free entry into the business, sug-gests that this form of competition is likely to be envi-ronmentally damaging. But it should also be notedthat, in Lima at least, free entry vastly improved ser-vice availability in comparison with that provided bythe former parastatal monopolist. Hence an interme-diate form of controlled competition for franchises, re-quiring a substantial public sector planning and pro-curement effort, is likely to yield the bestenvironmental outcome, as well as be economicallyefficient. Cities such as Bogotá have achieved signifi-cant improvements in service quality and reductionsin air pollution by buses along the TransMilenio corri-dors without setting extremely stringent technicalstandards. The lesson appears to be that attempts to

reduce pollution by the bus sector need to be embod-

ied in a comprehensive sector policy—one that usu-

ally requires substantial regulatory reform.

Where a certain decision is bound to have an ad-verse impact on the private sector, working closelywith the affected parties prior to implementing thedecision and possibly negotiating compensating mea-sures could go a long way in minimizing social un-rest. In Dhaka, all two-stroke engine three-wheelerswere banned as of January 2003. Given the largenumber of vehicle owners and drivers involved,numbering tens of thousands, this could have re-sulted in a large-scale general strike with serious eco-nomic and security consequences. However, years ofconsultation and negotiation prior to the implementa-tion of the ban—so that the affected individuals hadtime to consider how the new policy would affectthem and to seek alternative employment—helpedsmooth the transition.

Nongovernmental Organizations andCivil SocietyNearly all of civil society is affected by air qualitymanagement. Tighter standards usually mean moreexpensive fuels or vehicles or both. Emissions inspec-

tion means that vehicle owners have to report to testcenters, pay a fee to be tested, and pay to repair theirvehicles should they fail the test. Private car restraintwill inconvenience car owners.

Sector reforms are often prevented because asmall number of vested interest groups stand to ben-efit from the existing system. Informing the publicabout the advantages of sector reforms, including en-vironmental benefits, can create public pressure onthe government to press on with the reform measures.This is especially relevant when environmental im-provement is lagging behind because of protection ofdomestic industries that are inefficient or are so pow-erful that the government refrains from taking stepsthat will offend them. A classic example is a mo-nopoly state oil company resisting gasoline leadelimination. Tremendous public pressure can make iteasier for the segment of the government wanting tomove forward with tighter environmental standardsto confront powerful lobbies defending their own in-terests. Broader sector policy reforms, often essential

for the introduction of cleaner vehicles and fuels,

bring great benefits to society at large.

Good information is important. Government cando more to inform civil society about why it is takingcertain unpopular actions to reduce air pollution.Conversely, civil society, and particularly NGOs, canbring pressure on government to take more action byraising public awareness and conducting informationcampaigns. For example, the decision to increase thetax on diesel to better reflect its externality (road, con-gestion, and environmental damage caused by theuse of diesel not paid for by the diesel user) may besound but is almost universally politically unpopular.It would be helpful to inform the public about thedownside of not increasing the price of diesel so thatresidents are fully informed about the costs as well asthe benefits of having cheap diesel. More generally,NGOs and the media can be especially helpful in rais-ing public awareness about the costs and benefits ofmeasures to improve air quality, especially where be-havioral changes are involved. Campaigns to raiseawareness about smoke emissions, proper vehiclemaintenance or operation, and air quality have beensuccessfully conducted in a number of developing


country cities. NGOs and the media can urge and help

the government implement new policy measures by

playing an advocacy role.

To this end, removing institutional and marketbarriers to the involvement of NGOs and the privatesector in areas such as fuel quality monitoring and ve-hicle inspection and maintenance programs would bea useful first step. There is already extensive evidenceon how to avoid corruption and increase efficiency invehicle I/M in Mexico. However, it is also importantto stress that outsourcing these functions does notmean that the cost to the government will fall dra-matically. The experience in Mexico City demon-strates that the government must be committed to in-vesting the resources, staff, and effort in auditing andsupervising the environmentally driven programs toachieve high levels of objectivity and transparency. Inthis regard, having test stations in the hands of theprivate sector would make it much easier for the gov-ernment to supervise and monitor the stations. It ismore difficult for one government agency to applysanctions to another government agency (for ex-ample, by taking away the license to operate test cen-ters or firing staff found to be engaging in corruption)than it would be to take this step against a private sec-tor contractor. The potential for NGOs and the pri-

vate sector to take over from the government some

monitoring and enforcement responsibilities should

be explored.

Enforcing compliance is important not only fromthe point of view of air quality improvement, butmore importantly, for promoting sector efficiency.Otherwise, a likely outcome is partial or total degra-dation of the market—an example of Gresham’s Lawin which the low-quality product drives out the high-quality product because of an inability to distinguish

Campaigns to educate the public about proper vehicle maintenance

and use of lubricant for two-stroke engines, such as one organized

by motorcycle manufacturers in Bangkok shown here, promote cost-

effective mitigation measures.

between the two. This applies not only to fuels, spareparts, and other items that consumers buy directly,but also to whether a factory is equipped to meetemission and discharge standards, or a goods carriercan meet the safety and exhaust emission standards.The legitimate private sector can benefit from betterstandards of compliance. This may arise when thelevel of public awareness about frauds in fuel or autopart markets increases, so that those who guaranteequality can expand their market share (see Interna-tional Experience 3). This underscores the importanceof public education. “Naming and shaming,”whereby those in fragrant violation of environmentalregulations are published by name, can be another ef-fective strategy, as is public listing of “green” compa-nies and products. Creating market conditions that

would encourage private sector actors to police them-

selves for compliance with regulations and standards

is a “no regret” situation.

85

Synopsis: Constructing an EffectivePackage of Measures

Adopting a Positive Policy StanceThe transport sector has contributed to the deteriora-tion of air quality in many large cities worldwide. Butin a number of industrial country cities, declining airquality has been halted or reversed. Initially that im-provement was associated with the elimination ofmajor pollution from stationary sources, phasing outthe use of coal in a number of coal-consuming cities,and more recently by addressing air pollution fromthe transport sector where the introduction of cleanervehicle and fuel technology has resulted in dramati-cally lower emissions. Traffic demand managementmeasures, bus priority measures, and the promotionof cycling have also played a role in reducing air pol-lution in a growing number of cities in both devel-oped and developing countries. In some richer devel-oping country cities the situation is not far removedfrom the situation of the industrial countries, wherethe value to society of improving air quality has in-creased faster than the marginal cost of advancedtechnologies. In those cities—for example, Santiago—investment in state-of-the-art fuel and vehicle tech-nologies can be a logical and cost-effective step in airquality improvement.

Cost-effectiveness is a key consideration for devel-oping countries as they devise their air quality poli-cies. It should be remembered that only after adoptinglower-cost options have industrial countries consid-ered more costly mitigation measures requiring theuse of emerging technology. In fact, the significant im-provements in reduction in pollution from mobilesources that have allowed most industrial countries tomeet the air quality guidelines recommended by theWorld Health Organization have come through well-established technological and operational strategies

that are discussed in this report. The most advancedstate-of-the-art technologies are still waiting to be in-troduced on a large scale even in the high-income in-dustrial countries. Developing countries moving di-rectly to more costly measures for controlling airpollution may simply divert resources from invest-ments that offer greater reduction in environmentalpollution or incur significantly more expenses toachieve smaller additional reductions in absoluteterms.

The most aggressive and bold actions to controltransport-related emissions should be undertaken inthose cities with the most serious air quality prob-lems and where the transport sector is a major con-tributor. However, even where transport’s contribu-tion is not currently high—such as in majorcoal-consuming countries or in low-income citieswith a high percentage of solid fuel use—there is noexcuse for taking a resigned attitude to air pollutionfrom mobile sources. It is wise to anticipate the in-crease in transport emissions that has accompaniedeconomic growth in every municipality worldwide.

Because of the great variety of national situationsthere is no universally applicable strategy for optimalreduction of transport sector emissions. Rather, it isnecessary for decisionmakers to consider policieswithin their own technical, economic, political, andinstitutional circumstances.

The policy tools at their disposal can be broadlyclassified into two types:

� Direct policy tools that specifically target airquality improvement

� Indirect policy tools with objectives other thanair quality improvement, but where there arecollateral benefits in environmental improve-

C H A P T E R8


ment, economic growth, and long-term povertyreduction.

Direct Policy ToolsDirect policy tools have air quality improvement astheir primary objective. Some have very low cost-to-benefit ratios (“low-hanging fruit”), and hence sub-stantial robustness to differences in national situations.

Targeting the greatest dangers to healthThe initial targets for action in developing countriesshould be those pollutants that are known to causethe greatest damage to health:

� Lead is strictly a matter of gasoline specification,and the move to ban lead in gasoline is effective.Stopping the addition of lead to gasoline, whichcan be done at a relatively low cost, instantlystops lead emissions from all gasoline-fueled ve-hicles, regardless of their age or state of repair.

� Fine particulate matter originating especiallyfrom diesel-powered vehicles and two-strokeengine two- and three-wheelers, is a seriousproblem in developing countries.

� Ozone precursor control is a high priority in alimited but growing number of cities, often withspecific climatic conditions.

Gross polluters

Worldwide, much of the emphasis has been on theemission characteristics of new vehicles. Even in theUnited States, however, a study found that poorlymaintained vehicles, representing 20 percent of all ve-hicles on the road, contributed about 80 percent of to-tal vehicular emissions (Auto/Oil Air Quality Im-provement Research Program 1997). The problem ofold and poorly maintained vehicles is even moretransparently obvious in most developing countries.Dealing with these requires:

� Efficient identification of target groups (usuallyold, high mileage trucks and buses)

� Development of instruments for their testingand repair

� In some circumstances, assistance with prema-ture scrapping.

Fuels and fuel standardsThe appropriate standards for fuel will depend oncountry circumstances, including the level of air pol-lution and the costs of upgrading. But some generalguidelines can be stated. The first three steps are ex-pected to have high benefit-to-cost ratios:

� The first priority should be to move to unleadedgasoline while ensuring that benzene and totalaromatics do not rise to unacceptable levels.

� Sulfur in gasoline should be lowered to a maxi-mum of 500 wt ppm and preferably lower assoon as possible to enable efficient operation ofcatalytic converters (following lead removal).

� Where sulfur content in diesel is very high, astrategy to lower it to 500 wt ppm or lowershould be identified and implemented. Whencombined with modern engine technology andgood maintenance practice, sulfur reduction to500 wt ppm would enable significant reductionsin particulate emissions. Where moving to 500wt ppm is very difficult in the near term butlowering sulfur to 2,000–3,000 wt ppm is rela-tively inexpensive, efforts should be made tomove to this level immediately.

� Where the resource and infrastructure condi-tions for natural gas are favorable and those forclean diesel technology are much less so, consid-eration may be given to shifting high mileagepublic transport fleets from diesel to CNG.

� In countries with current or potentially high lev-els of air pollution from mobile sources, espe-cially those that have already taken steps to-ward 500 wt ppm, or where new oil refiningcapacity is being added or major renovationsbeing undertaken, the cost-effectiveness of mov-ing to ultralow sulfur standards should be ex-amined, taking into account maintenance capa-bility and the concomitant investments in thenecessary emission control technologies to ex-ploit lower sulfur fuels.

CHAPTER 8. SYNOPSIS: CONSTRUCTING AN EFFECTIVE PACKAGE OF MEASURES 87

Vehicles and vehicle standardsThe setting of appropriate standards for vehicles is anessential complement to the determination of fuelstandards. General guidelines for setting these stan-dards are listed below. The third and fourth recom-mendations for motorcycles and catalytic convertersare likely to have high benefit-to-cost ratios.

� Emission standards for in-use vehicles shouldbe set at levels that are achievable by a majorityof vehicles with good maintenance, and shouldbe tightened over time.

� Vehicle emission standards for new vehiclesshould be progressively tightened to levels con-sistent with improving fuel quality.

� All new two-stroke engine motorcycles shouldbe required to meet the same emission stan-dards as four-stroke motorcycles.

� The installation and continued maintenance ofcatalytic converters should be required for allnew gasoline-powered vehicles in countrieswhere lead in gasoline has been eliminated.

� The introduction of particulate filters and otherdevices to reduce end-of-pipe emissions fromdiesel vehicles should be considered whereultralow sulfur diesel is available. As filter andother device technology develops and pricesfall, and as low-sulfur fuels become more widelyavailable, this strategy will become more robust.

Inspection and maintenanceA targeted and effective vehicle inspection program isnecessary to make vehicle standards effective. Inter-national experience suggests the following advice onestablishment of an I/M system:

� Centralized, test-only private sector centers withmodern instrumentation, maximum automa-tion, and “blind test” procedures, and subject toindependent monitoring are most effective.

� In some circumstances, targeted incentiveschemes for scrappage or replacement of highmileage gross polluters may be considered forcomplementing I/M.

� Education campaigns and clinics should be un-dertaken to improve two-stroke engine mainte-nance and lubrication.

Urban designMuch suspended particulate matter in developingcountry cities comes from re-suspension of road dust.This can often be addressed by simple urban designsand landscaping that can be implemented at a smallincremental cost in many road projects, bringing sig-nificant environmental benefits. Whenever possible,road projects should include provision for paving allsections of the road including sidewalks, and wherepaving is not practical, landscaping with trees that re-quire no watering.

Institutional developmentEnsuring that transport and environmental policiesare consistent and well integrated requires an appro-priate institutional basis. A basis specifically targetingair quality improvement requires:

� An effective air quality monitoring regime be es-tablished in large cities, with institutional re-sponsibility clearly assigned and with adequateresources allocated.

� Institutions for administering and enforcing ve-hicle emission standards be established with aprimary task of identifying and removing fromthe road gross-polluting vehicles.

Legal sanctionsWhere regulatory measures and fines are insufficientto curb the worst offenders of local air quality, legalsanctions should be developed and implemented to:

� Prevent fuel adulteration and the smuggling oflow-quality fuels from neighboring countries.Consideration may be given to hold fuel mar-keters legally responsible for quality of fuelssold.

� Prevent the import of grossly polluting vehicles.� Ensure that testing stations are following correct

procedures.


Indirect Policy ToolsSome policy measures are indirect policy tools withprimary objectives other than air quality improve-ment, but which can give sizeable collateral air qual-ity benefits. They tend to have long “gestation” peri-ods, but can derail air quality management in the longrun if poorly handled.

Sector reformMany of the measures needed to reduce transport-re-lated air pollution will be suboptimal or ineffectiveover the longer term without the reform of the trans-port or fuel sector. The oversupply of bus or taxifleets, the misallocation of clean fuels such as naturalgas, or simply poor quality gasoline and diesel fuelcan all be symptoms of market distortions or regu-lated inefficiencies. Therefore, prior to or in parallelwith instituting regulatory and administrative mea-sures for reducing transport-related emissions, citiesand countries should ensure that transport and fuelsectors are efficiently (and equitably) organized. Afew examples of where sector reform can produce im-provements in air quality are described below:

� Rationalize public and private transport fleets,reduce predatory behavior by paratransit, andensure that private road users are not dispropor-tionately favored over public transportation. Tothe extent that traffic congestion is a result of theorganization of the transport sector, a city canmake major improvements in air quality by im-proving the efficiency of the sector.

� Reduce direct and indirect subsidies and protec-tion of domestic petroleum refineries and re-quire that they meet increasingly stringent fuelquality standards. Rather than investing largesums of capital in inefficient and unprofitablerefineries, the sector should be opened to newentry with the long-term objective of creating anopen and competitive market. Allowing im-ported fuels to compete with domestic refineriesis a particularly effective way of forcing effi-ciency improvement in the sector.

� Establish and enforce transparent and clearlydefined regulations. Lack of enforcement is an-

other way of protecting inefficient operators.Rigorous enforcement will also reduce the in-centives for smuggling of fuels and fuel adul-teration which are a major obstacle to maintain-ing fuel quality and reducing vehicle emissionsin developing countries.

Where prices are controlled by the government,unsustainable subsidies that discourage uptake ofcleaner fuels should be phased out. While sector re-forms cannot be the sole solution for reducing emis-sions from mobile sources, the efficiency of the trans-port and energy sectors must be a concern forenvironmental policymaking. Some of these reforms,such as import liberalization for clean fuels, will benational in scope, whereas others, such as urbantransport sector reform, will be local. In both cases,they are likely to have significant economic and socialbenefits aside from their environmental benefits, andin this sense should be seen as “no regret” measures.

Appropriate fiscal policiesFuel taxes should be used to reduce private vehicleuse, encourage fuel-efficient vehicles and the use ofcleaner fuels, and compensate for road wear and tearand environmental damages. The main implicationsof this are as follows:

� In many countries this will mean raising taxeson diesel fuel for transport use.

� In addition to fuel taxes, separate vehiclecharges should be considered based on vehicleweight, axel-loadings, and annual mileage.

� Direct charges for the use of urban road spaceshould be introduced, including congestioncharges.

� Taxes, import duties, and vehicle licensing dis-incentives should be introduced for pollutingvehicles and engines.

� Serious consideration should be given to elimi-nating subsidies to public off-street parking aswell as not permitting free on-street parking, es-pecially where they increase congestion by gen-erating private transport trips to congested loca-tions, or where on-street parking increasescongestion by reducing available road space.


Integration in transport planning andmanagement policiesAll of the technological improvements need to be setwithin a favorable transport planning and manage-ment policy framework. That framework requires thefollowing:

� Transit-oriented development strategies shouldbe developed to reduce trip lengths and concen-trate movements on efficient public transportaxial routes.

� Air quality audits of all new major transport in-frastructure projects should be undertaken aspart of environmental impact assessment proce-dures to determine if projects will lead to orworsen exceedances of ambient air quality stan-dards.

� Priority should be given to buses in the use ofroad infrastructure, and particularly the creationof segregated busway systems, in order to im-prove and sustain environmental standards forbuses.

� The efficiency of bus operation should be im-proved through the design of more efficientroute networks, better cost control, and by thecreation of incentives for improvement throughcommercialization and competition.

� Adequate pedestrian and bicycle facilitiesshould be established in order to promotenonmotorized options for short distance trips.

� Protocols for traffic signal system settingsshould be established and implemented whichresult in reduced air quality emissions withoutcompromising pedestrian and bicyclist safety.

� Fiscal and/or administrative devices should beimplemented for restraining private-car trafficin congested areas, particularly at peak times.

� Competitive bidding for transport franchisesshould be promoted based on performance-based criteria, including emission characteristicsof vehicles.

� Urban traffic management centers should be es-tablished, involving police in traffic manage-ment system design and training.

� A municipal department or agency with com-prehensive responsibility for integrated land useand transport planning, including environmen-tal protection issues, should be created.

Political and Technical ConsistencyExperience suggests that strategies for reduction ofurban air pollution are most likely to be successfulwhere they are internally consistent in a technicalsense, and as far as possible consistent with other sec-tor strategies.

Technical consistencyRegardless of what vehicle emission and fuel qualitystandards are set, they should be technically consis-tent. The following are a few real world exampleswhere technical consistency has not been followed:

� Mandating catalytic converters for gasoline ve-hicles in the absence of a reliable supply of un-leaded gasoline.

� Mandating oxidation catalysts in buses whendiesel available on the market contains rela-tively high levels of sulfur (significantly exceed-ing 500 wt ppm), which not only leads to rapidcatalyst deactivation but also increases particu-late emissions markedly by facilitating the oxi-dation of sulfur to sulfate while the catalyst isstill effective.

� Mandating particulate filters in all diesel ve-hicles when the sulfur content of diesel sold onthe market is in the thousands of wt ppm, inca-pable of enabling effective operation of the filter.

� Setting the same emission standards for vehiclesirrespective of their weight, so that heavy trucksare required to meet the same particulate emis-sion standards in g/km as small passenger cars.

� Failing to establish sufficient refueling capacitybefore or at the same time as mandating a large-scale conversion to CNG.

An audit on the technical consistency of a strat-

egy, including consideration of upstream (fuel and

other infrastructure) and downstream (transport sec-


tor service conditions) interactions is an essential re-

quirement of an urban air quality strategy.

Avoiding conflict with transport economyUltimately, policies for air quality have to be politi-cally feasible. This means that they have to be seen asnot inimical to the interests of those capable of build-ing a political consensus against them. A very com-mon conflict is that with economy of transport. Forexample, a major source of particulate pollution islikely to be heavy diesel engine trucks and buses. Inthose circumstances simply forcing higher-cost ve-hicles on to the formal public transport sector(whether gas or modern diesel) may have the per-verse effect of causing that sector to be replaced, le-gally or illegally, by a fragmented, smaller-vehicle in-formal sector that may be no less polluting but muchmore difficult to control. To avoid this, a multi-strandpolicy is likely to be required. Modernization of fleetsthrough assisted scrappage programs, served by ef-fective inspection programs to identify the appropri-ate vehicles for replacement, may be effective. Inter-national experience has demonstrated that attempts

to improve the quality of the public transport fleet

are most likely to be effective if undertaken in the

context of a thorough reform of regulation of the sec-

tor.

Social acceptabilityAn even more difficult conflict arises where existinghigh levels of personal mobility are constrained in or-der to reduce air pollution. For example, if pollutionarises from privately owned two-stroke engine motor-cycles (typical of many Asian cities), the most effec-tive short-term instruments are likely to be programsto improve lubrication behavior. For the mediumterm, the most effective response would be to tightenemission standards, leading to significant replace-ment of two-stroke engines by four-stroke engines (aswell as introduction of new direct injection two-strokeengine technology). In the very long term, however,this is not likely to be a sustainable solution, becausethe motorcycles are replaced by motor cars as incomerises, and congestion increases. Hence, even in citiessuch as Hanoi or Ho Chi Minh City in Vietnam,

which have high levels of four-stroke engine motor-cycle ownership and very high levels of personal mo-bility, further long-term action will be needed to man-age and restrain car use. Such long-term action maycall for a broad policy, including traffic managementand restraint through parking and other policies, to-gether with public transport development. Thatwould include bus priorities and segregation, or pos-sibly even rail system development in the largest cit-ies. Those types of investment, while environmentallybeneficial, will probably also be justified by transport-efficiency considerations.

“Horses for Courses”The policy stance adopted in this report is that strate-gies need to be adapted to local problems and possi-bilities. Mostly that impinges on the pace at whichstate-of-the-art technology can be adopted as incomesincrease. But some other characteristics of cities alsoinfluence the selection of strategies.

Cities in fuel-producing countriesCuriously, cities in fuel-producing countries may faceparticular difficulties with fuel quality. This is becausethey may perceive a need to protect the local manu-facturers and be tempted to do this by a liberal ap-proach to fuel quality. This is a particular problem incountries with government-owned refineries, wherefinance ministries may be resistant to the very largeinvestments necessary to improve fuel quality (par-ticularly to bring down sulfur content). They may alsoresist competition from cleaner imported fuels for bal-ance-of-payments reasons. It is not uncommon to finda politically powerful and connected monopoly stateoil company resolutely opposing tightening of fuelstandards, even something as straightforward asgasoline lead elimination.

The solution is to set what are considered to beeconomically and environmentally supportable fuelstandards and to open the market to imported as wellas domestically refined fuels. But there is no easy pathto this solution. What is worth doing, however, is toundertake a study of achieving different levels ofstandards and to compare costs of imports with those


of domestically refined fuels. If the cost of the formeris markedly lower, it is important to understand whyand inform the public openly so that benefits andcosts of protecting the domestic oil industry can befully appreciated.

If the country is fortunate enough to have an in-digenous supply of natural gas and an existing distri-bution network, a program of assisted fuel switchingto CNG in heavy vehicles may be a consideration.Usually, however, CNG is not markedly cheaper thandiesel as a transport fuel, so that the cost-effectivenessof the use of CNG is a major problem. The structureand levels of fuel taxation may also need to be ad-justed to give the appropriate incentives to operategas vehicles without seriously depleting tax revenuesor simply generating a replacement of gasoline, ratherthan diesel, vehicles by natural gas. Even for a coun-try with gas resources and city distribution, such asColombia, the distribution costs from gas field to mar-ket may leave the full distributed cost of gasuncompetitive with that of liquid fuel, so that sup-porting tax discrimination will still be necessary forlarge-scale switching from diesel to CNG. The appar-

ent balance-of-payments benefit of using a domestic

resource needs to be carefully weighed against the

sector efficiency and macroeconomic costs of subsidy

to domestic fuel producers.

Cities in vehicle-manufacturing countriesVehicle-manufacturing capacity is unlikely to cause acountry any problem if it is privately owned, subjectto competition from imports, and at least in part de-pendent on exports for its livelihood. In those circum-stances vehicle quality is unlikely to be a big problemas companies will need to meet standards that wouldbe acceptable in industrial-country markets. That hasbeen the case in emerging manufacturing countriessuch as the Republic of Korea. However, it is not un-usual in these cases for the auto industry to be pittedagainst an oil industry that does not want to bear thecosts of manufacturing fuels to the higher standard.Government needs to mediate and arrive at a com-promise position.

The presence of vehicle-manufacturing capacitymay also make vehicle I/M easier to the extent that

there is likely to be greater local technical capacity tohandle I/M. In non-manufacturing countries all ve-hicles are imported. The appropriate stance thenwould seem to be to adopt standards related to thoseof the source country that can be achieved by prop-erly maintained vehicles. That would to some extentbe hampered because local technical expertise in ve-hicle I/M would likely be lower, making enforcementof standards more difficult.

The greatest danger in developing countries thatmanufacture vehicles is likely to arise from protectivetrading policies. There will be a temptation to pro-scribe, or severely tax, the import of new vehiclescompeting with those manufactured domestically. Atthe same time, because of a perceived need forcheaper, secondhand vehicles, in some countries re-strictions on the import of used vehicles are likely tobe nonexistent or much less stringent. This is likely tohave the perverse effect of encouraging the import ofolder, lower-standard vehicles.

The role of economic philosophy: the commandeconomiesSome of the most severe problems of transport-gener-ated air pollution arise in the command economiesand the transition economies. In such economies, for-mal sector public transport provision, fuel pricing,and vehicle imports tend to be tightly controlled bythe government. But the disappearance of the formersources of support for public transport in the transi-tion economies and the increasing difficulty of main-taining it in some of the remaining controlled econo-mies (for example, the central Asian republics of theformer Soviet Union) have led to a collapse of the tra-ditional large vehicle public transport services onwhich the population was primarily dependent tomeet its movement needs. Public transport vehiclesare not being adequately maintained or replaced,with the effect that those still on the road are highlyinefficient and polluting by industrial country stan-dards. The disappearing traditional service tends tobe supplemented or replaced by informal sector ser-vices operating under inefficient, quasilegal regula-tory arrangements, often with inappropriate vehicles.For example, public transport in the secondary cities


of Uzbekistan is increasingly dependent on the seven-seater Daewoo Damas minivan.

Sector reform is crucial in those countries. Indeed,significant progress in urban air quality managementis unlikely unless sector reforms are carried out inparallel with imposition and enforcement of emissionstandards. The agenda for such countries thus ap-pears to be that of leveraging private capital, promot-ing market-based solutions, and undertaking publiceducation to create market conditions favorable toemissions reductions.

Some special cases?Some cities will inevitably not be adequately compre-hended in this limited taxonomy. For example, manyhistoric cities do not have enough space to carry thetraffic volumes generated or to provide segregatedpublic transport systems. For those cities, the experi-ence of some of the European cities such as Zurichmay offer the most appropriate model. In that andmany other cases, some locations have been closedcompletely to motorized traffic; in others entry is lim-ited to public transport. The success of some of theEuropean cities in maintaining public transport accessin constricted space can be replicated in developingcountries if the political will exists to give adequatepriority to public transport in the allocation of space.

At the other extreme are cities that have been de-veloped at artificially low densities, such as the cities

of apartheid South Africa. On the one hand, these cit-ies present enormous social problems; on the other,they offer the possibility for innovative planning solu-tions in the process of restructuring land use. Landuse planning and the allocation of adequate space forsegregated public transport is particularly importantto air quality in those cities.

ConclusionWhile the specific actions for reducing transport emis-sions will vary from city to city, there are several un-derlying principles that this report seeks to emphasizefor building an effective policy package:

� Raise awareness among policymakers and thegeneral public about urban air pollution levelsand damages and specify and promote the rolethat transport plays.

� Press for sector reform that increases sector effi-ciency, benefits society at large by providinggoods and services at lower cost, and at thesame time reduces emissions.

� Raise awareness in business as well with con-sumers about business “best practice” that isalso likely to bring about environmental benefitsto society.

� Work with, not against, the economic incentivesof various transport actors.

93

Conventional Fuel Technology

The majority of vehicles worldwide will continue touse gasoline and diesel for the foreseeable future. Thisannex focuses on fuel and vehicle technology involv-ing these two liquid fuels. It discusses fuel quality im-provement and what is driving the evolution of fuelspecifications. The specific fuel and vehicle standardsare treated in annexes 2, 3, and 4. (Two- and three-wheelers are treated in annex 5.)

Gasoline and diesel fuels are made up primarily ofhydrocarbons. Virtually all components are naturallyfound in these fuels with the exceptions of olefins(which are formed in refining processes) and lead,oxygenates, and detergents (which are externallyadded). The primary product of combustion is CO2, agreenhouse gas with no adverse health effects. Othercombustion products include CO and hydrocarbons,both products of incomplete combustion; NOx, whichis formed when nitrogen in air is subjected to hightemperature in the presence of oxygen in the combus-tion chamber; and various decomposition (for ex-ample, benzene from nonbenzene aromatics) or poly-merization, dehydrocyclization (for example, PAHs),and dehydrogenation (for example, soot) products.Some fuel parameters affect emissions of all vehiclesessentially equally. For example, the amount of leademitted is approximately proportional to the amountof lead contained in gasoline multiplied by theamount of gasoline consumed. For the most part,however, pollutant emission levels are determinednot only by the fuel composition and the engine typebut also by specific engine and driving characteristicsthat interact in complex ways. A given vehicle usingthe same fuel will show different emission levels de-pending on the driving cycle.

There have been enormous changes in fuel specifi-cations worldwide in the last three decades. Thesechanges can be categorized into two groups:

A N N E X1

� Those leading directly to reductions in exhaustor evaporative emissions. Beginning with leadphase-out, these steps have historically beentaken in stages over a number of years.

� More recently, those for enabling the market in-troduction of very advanced emission controltechnology. The dramatic reductions in fuel sul-fur envisaged in Europe and North Americaduring the latter half of this decade are primeexamples of this.

Gasoline Quality ImprovementGasoline is a volatile fuel, and hence evaporativeemissions of hydrocarbons need to be taken into ac-count in addition to exhaust emissions. Properlytuned gasoline vehicles typically emit far less particu-late matter in mass than conventional diesel technol-ogy vehicles. That said, gasoline vehicles may havehigh number counts of particles, and number countsmay be of interest in the future. The gasoline fuel pa-rameters that affect air quality are listed in table A1.1.

The extent to which undesirable fuel componentscan be reduced is primarily a question of the costs ofremoving them—for example, dramatic reductions inbenzene and total aromatics or production of sulfur-free gasoline would be expensive.

Benzene and total aromatics tend to increase ini-tially as lead is phased out to make up for the octaneshortfall. Benzene is a carcinogen and is found ingasoline as well as formed as a by-product of combus-tion of other aromatics. Various studies have indi-cated the cost-effectiveness of controlling benzene it-self rather than total aromatics. Controlling totalaromatics, which are benzene and ozone precursors,is an important consideration but not as much as con-trolling benzene itself.


Gasoline Fuel Parameters that Affect Air Quality

Parameter Characteristic feature

Lead Lead is an inexpensive octane enhancer and has historically been added to gasoline as alkyl lead. It deactivatescatalytic converters permanently. Because lead is extremely toxic, its use in gasoline has been banned in agrowing number of countries.

Benzene Benzene, the smallest aromatic compound with no alkyl groups, is a carcinogen. It is emitted from gasoline as aresult of evaporation and as unconverted benzene from the exhaust pipe. Benzene has an extremely high octanerating, and is hence a good gasoline blending component from the point of view of combustion.

Sulfur Sulfur in gasoline acts as a poison for conventional catalytic converters. (The effect of the poison is temporary andreversible to a large extent.) Vehicle manufacturers recommend for conventional catalytic converters that the levelof sulfur in gasoline be kept below 500 wt ppm, and preferably below 100 wt ppm. The impact of reducing sulfuron the performance of catalytic converters follows a nonlinear relationship, with emissions decreasing morerapidly below 100 to 150 wt ppm (MECA 1998). Sulfur not trapped in catalytic converters is emitted as SO

x, some

of which undergoes chemical transformation to become secondary particulate matter or acid rain.

RVP The Reid vapor pressure is a measure of gasoline volatility. Lowering RVP is a cost-effective way of controllingVOC emissions, including the emissions of light olefins. This requires lowering the level of butanes, and, possibly,C

5 hydrocarbons (hydrocarbons with five carbon atoms). Butanes are the cheapest source of octane, and their

removal typically adversely affects refinery economics. Lowering RVP, however, can make starting carburetedvehicles difficult, particularly in cold weather, and may cause misfires and higher tailpipe HC emissions. RVPshould be within a certain range depending on vehicle technology and ambient conditions.

Aromatics Aromatics with two or more alkyl groups are photochemically reactive and contribute to ozone formation. Alkyl-aromatics (that is, nonbenzene aromatics) also dealkylate (lose alkyl groups) during combustion, and a fraction isemitted as benzene. However, it takes 10–20 times as much alkyl-aromatics to form benzene in the exhaust gasas benzene found in gasoline itself. The photochemical reactivity of aromatics and their decomposition to benzeneare the two primary environmental concerns with aromatics. Aromatics have extremely high octane ratings, andhence are good gasoline blending components from the point of view of combustion.

Olefins Olefins in gasoline are formed primarily during cracking processes. Olefins are photochemically reactive and areozone precursors. This is the primary concern. In addition, at elevated levels, olefins increase the emissions ofNO

x. NO

x is a precursor for ozone, particulate matter, and acid rain. Olefins have fairly high octane ratings, and

hence saturating them to reduce their quantity adversely affects the combustion characteristics of gasoline.

Alkylates Alkylates are high-octane paraffinic hydrocarbons with essentially no adverse effects on air quality. They areexcellent substitutes for less desirable blending components such as aromatics. However, alkylation is anexpensive process and requires the presence of a catalytic cracking unit at the refinery.

VOCs Volatile organic compounds contain photochemically reactive hydrocarbons. Reductions in VOC emissions willtherefore reduce the amount of ozone precursors in the atmosphere. VOCs could also adsorb onto particles andincrease particle mass. Evaporative emissions consist entirely of VOCs; VOCs are also found in exhaust gas.

Oxygen Oxygenates such as ethers and alcohols have high-blending octane and hence help compensate for octaneshortfall after lead removal. The presence of oxygen in oxygenates also facilitates clean combustion in vehiclesthat are not equipped with oxygen sensors and that are running rich (low air-to-fuel ratio). If a vehicle isreasonably adjusted, however, oxygenates could raise NO

x, and also cause lean misfire, raising HC emissions.

Oxygenates can increase the emissions of undesirable compounds such as aldehydes. Oxygenates also dilutegasoline, thereby decreasing the amount of such undesirable gasoline components as benzene, sulfur, totalaromatics, and olefins. Oxygenates are more miscible with water than gasoline, and contamination with groundand drinking water with MTBE, an ether, has been a concern in the United States.

Additives Gasoline in industrial countries contains additives to prevent the accumulation of deposits in engines and fuelsupply systems. Deposits can increase tailpipe emissions.

T A B L EA 1. 1

ANNEX 1. CONVENTIONAL FUEL TECHNOLOGY 95

Volatility control reduces hydrocarbon evapora-tive emissions and can be especially effective for ad-dressing ozone if the low boiling fraction of gasolinecontains a sizable amount of light olefins, which arestrong ozone precursors. In cold weather, lower vola-tility can cause problems with starting and accelerat-ing old carbureted vehicles and may cause a rise intailpipe discharge of unburned fuel. Fortunately,ozone is primarily a summertime problem in manycities, making volatility control important during thesummer. Lowering gasoline volatility, with carefulseasonal management of volatility control, merits seri-ous consideration in cities where ozone is a concern.

In addition, in order to take advantage of thegrowing number of catalyst-equipped cars, sulfur re-duction should be seriously considered in countrieswith gasoline containing high levels of sulfur. Sulfurreduces the efficiency of the catalyst, and decreasingsulfur has a large impact on reducing exhaust emis-sions. Phosphorus is another catalyst poison, so addi-tives containing phosphorus should be restricted.

In contrast, lead has traditionally been added togasoline as an octane booster to ensure smooth com-bustion. In the 1970s, little was known about the tox-icity of lead, but it was known that lead in gasolinedeactivated catalytic converters. The initial reason forlead removal was thus as an enabling specification to fa-cilitate the installation of catalytic converters in cars.Only in the 1980s did there begin to emerge mountingevidence on the adverse health impact of lead expo-sure, even at levels previously considered safe, andparticularly to young children.1 The first step in gaso-line reformulation is therefore typically to stop the ad-dition of lead. A crucial issue in lead elimination is se-lecting alternative octane sources in a way thatminimizes their adverse effects on air quality. Devel-oping countries that are now phasing lead out of

gasoline have the advantage of learning from the ex-perience of countries that have already done so.

Diesel Quality ImprovementDiesel combustion gives rise to high levels of smallparticles and NOx. Smoke in diesel exhaust is causedby incomplete combustion. White smoke is caused bytiny droplets of unburned diesel resulting from en-gine misfiring at low temperature. This smoke shoulddisappear as the engine warms up. Black smokecould be caused by a faulty injector, insufficient air,underpowering or overfueling the engine, or both.Blue-gray smoke is the result of burning lubricatingoil and is an indication that the engine is in poor me-chanical condition.

Vehicle exhaust particulate emissions consist ofcarbonaceous and sulfate-based components. Becausevehicle technology improvement addresses carbon-aceous components only, in the United States sulfatecontributions to particulate emissions, which are afunction only of diesel sulfur content, became signifi-cant by 1993 in percentage terms, and hence the needto reduce sulfur in diesel to 500 wt ppm. Evolvingcontributions of carbonaceous and sulfate compo-nents to particulate emissions are illustrated in figureA1.1. It is clear from the figure that by the 1994 modelyear, the vehicle technology improvement for mini-mizing particulate emissions reached the point where

1Countries that have already banned lead in gasoline includeBrazil (1991), the Slovak Republic (1996), Thailand (1996), Mexico(1997), Hungary (1999), Bangladesh (1999), the Philippines (2000),Georgia (2000), India (2000), Vietnam (2001), Sri Lanka (2002), anda host of Central and South American countries beginning in theearly 1990s. But some large countries, including Indonesia, Ven-ezuela, and a number of countries in Sub-Saharan Africa, stillhave leaded fuel.

Particulate Emissions from New U.S. Heavy-Duty Diesels

F I G U R EA 1. 1

Source: McCarthy 1994.

0

0.1

0.2

0.3

0.4

0.5

0.6

1988 1991 1994 1994

g PM/horsepower-hour

Carbon soot & organicsSulfate

Model year

0.25% sulfur

0.05% sulfur


Diesel Parameters that Affect Air Quality

Parameter Characteristic feature

Density Diesel fuel is metered volumetrically, so that the higher the density, the more mass is injected. The use of a fuelwith greater density than that used in the pump calibration could even result in overfueling at maximum load,resulting in substantially higher smoke emissions.

Sulfur Sulfur in diesel contributes to secondary particulate formation, elevated ambient concentrations of SO2, and acid

rain. It also decreases the efficiency of oxidation catalysts, increases sulfate-based particulate emissions while thecatalyst is active, and degrades advanced exhaust treatment devices (such as particulate filters) rapidly.

Aromatics Polyaromatic hydrocarbons (aromatics with more than one benzene ring) have been linked to higher particulateemissions. The impact of reducing aromatics on vehicular emissions is less clear. Although reducing aromatics isoften linked to lower emissions, a cooperative program between Esso and Statoil found, for example, that reducingtotal aromatics from 32 percent to 10 percent had no marked effect on particulate emissions (Betts and others1992).

T90

, T95

T90

and T95

are the temperatures at which 90 and 95 percent, respectively, of diesel evaporates. Decreasing T90

orT

95 could have a favorable impact on emissions, especially if heavy PAHs are removed as a result. Findings on the

impact of varying T95

are discussed further under the European Auto-Oil Programme in annex 3.

Cetane Cetane is a measure of the ignition quality of a diesel fuel. A high cetane number indicates a shorter lag betweenfuel injection and ignition. The cetane number is measured in an engine test (based on an outmoded engine thatdoes not match modern engines). A cetane index is an approximation of the cetane number calculated from thedensity and distillation temperatures. Cetane improvement additives will increase the cetane number but not thecetane index, and hence many countries set cetane specifications as minimal cetane index or cetane number.

Cleanliness With respect to satisfactory operation of diesel vehicles, cleanliness refers to the absence of water and particulatecontamination. Dirt and water can plug fuel filters and cause serious damage to the fuel injection system becauseof the close tolerance of fuel pumps and injectors. Diesel engines are equipped with fuel filters to protect the fueldelivery system.

Stability Stability is the ability of a fuel to resist the formation of gums and insoluble oxidation products. Fuels with pooroxidation stability contain insoluble particles that can plug fuel filters, potentially leading to decreased engineperformance or engine stalling from fuel starvation.

T A B L EA 1. 2

it would not have been cost-effective, or even pos-sible, to reduce particulate emissions by means of fur-ther vehicle technology advances only.

Many developing countries have relatively highT90 or T95 so as to maximize the yield of diesel. Thistends to exacerbate the problem of particulate air pol-lution because heavy diesel “ends” are more difficultto combust. Lowering density (which, broadly speak-ing, would mean lowering the back-end of the distil-lation curve) makes the fuel-air mixture becomeleaner, since diesel engines inject a constant volume offuel into a fixed amount of air. Leaner mixtures lowerparticulate emissions.

In order to reduce particulate emissions from die-sel engines, standard engine overhaul and mainte-nance, coupled with diesel quality improvement, can

have a significant impact. The environmental impactof various diesel parameters is listed in table A1.2.

The significance of sulfur in dieselMany policymakers in developing countries equatedramatic reductions in diesel sulfur with a markedimprovement in air quality. This arises in part fromobserving the heated debate that occurred betweenthe auto and oil industries in Europe and NorthAmerica about whether or not it would be cost-effec-tive to mandate “sulfur-free” (10–15 wt ppm sulfur orlower) diesel to control particulate and NOx emis-sions. For the purpose of controlling particulate emis-sions, the composition of particles matters. If the ma-jority of particles are carbon and not sulfate based,lowering sulfur in diesel may do little to reduce par-


ticulate emissions. In industrial countries, vehicletechnology has been steadily improving so that par-ticulate and NOx emissions from new diesel vehiclesare already very low. The only way to lower theiremissions further would be to employ advanced ex-haust-control technologies, some still under develop-ment, such as continuously regenerating traps (theperformance of which may not be as good in passen-ger vehicle applications as that in commercial vehiclesbecause passenger vehicles use full power less) andNOx adsorbers (not yet fully proven on the road, andno tests demonstrating durability as yet) or reductioncatalysts that can work at high air-to-fuel ratios. Mostof them in turn are extremely sulfur-intolerant.

There are a number of developing countries wherethe limit on sulfur in diesel is as high as 10,000 wtppm (equivalent to 1 percent). In such countries, low-ering the diesel sulfur limit merits immediate atten-tion. In parallel, steps should be taken to lower car-bonaceous contributions (unrelated to fuel sulfur) toparticulate emissions. An important milestone in theearly stage of sulfur reduction is to lower diesel sulfurto 500 wt ppm, as the United States did in 1993 andthe European Union in 1996. Sulfur reduction downto 500 wt ppm or lower is likely to have a measurableimpact on particulate emissions from existing ve-hicles. This would also enable the use of oxidationcatalysts in diesel vehicles. Although oxidation cata-lysts do not deal with graphitic carbon, they do lowersoluble hydrocarbons attached to particles, loweringmass particulate emissions. However, oxidation cata-lysts oxidize sulfur, contributing to sulfate-based par-ticulate emissions, so that it is important to ensurethat suitable catalysts be selected from the point ofview of minimizing sulfate formation over the cata-lyst, or else particulate emissions can even increaserather than decrease. Another point that should beaddressed is that lowering sulfur levels in diesel fuelsfrom, for example, 3,000 ppm to 350 ppm reduces thelubricity of the fuel and can adversely affect the wearon high pressure injection pumps. When diesel fuelswith up to 500 wt ppm sulfur were first introduced insome markets, there was a rash of fuel injector O-ringfailures causing fuel system leaks. The failures were

limited to older vehicles, and the problems have notrecurred. The damage to seals caused by the use oflow sulfur fuels is an important issue that should beanticipated with an appropriate response prior toimplementing a large sulfur reduction.

Lowering sulfur in diesel can also have beneficialeffects on engine life and oil change intervals. Sulfurin the fuel burns in the combustion chamber and re-acts with water to form sulfurous and sulfuric acids.These acids cause corrosive wear on the engine. Theyalso degrade the oil and facilitate the formation ofgums and varnishes. Alkaline additives are intro-duced into the oil to neutralize acidic compounds,and the speed at which these alkaline additives areneutralized in turn depends on the fuel sulfur con-tent. A further contaminant is fuel soot from the in-complete combustion of fuel in the cooler parts of thecombustion chamber. Soot increases viscosity and alsocauses wear. Soot and water in the lubricant can pro-duce a sludge in the oil, resulting in lubrication prob-lems if the oil is not changed frequently.

The impact of lowering sulfur below 500 ppm onoil change intervals depends on what is governing theinterval: acid formation or soot in the oil. As an illus-trative example, the Department of Buses of the MTANew York City Transit has been operating 4,200 buseson 30 wt ppm sulfur diesel since September 2000. Asof April 2004, 2,800 of these buses were equippedwith particulate filters. Oil change intervals have beenshortened by the use of exhaust gas recirculation(EGR)—for NOx control—due to oil degradation fromgreater soot formation. One bus company in Finlandreported doubling of the oil change interval from10,000 km to 20,000 km when it switched from 500 wtppm sulfur diesel to reformulated diesel containingless than 50 wt ppm sulfur (Mikkonen and others1997). This would indicate that acid formation, andnot soot, was governing the oil change interval.Whether acidity or soot determines the oil change in-terval depends on the engine technology; modern,less polluting engines tend to force more soot throughthe oil than others. If the soot level is very much be-low the maximum level and the oil has to be changedbecause of acid formation, then diesel sulfur reduc-


tion is likely to lead to longer oil change intervals. Onthe other hand, if soot levels are near the maximumlimits, then a dramatic reduction in sulfur is unlikelyto lengthen the oil change interval. EGR increases sootformation significantly, so that the effect of soot fromEGR, rather than the acidification tendency of fuelsulfur, is likely to dominate. In industrial countrieswith diesel sulfur in the vicinity of 300–350 ppm, theamount of soot is the primary parameter governingoil change intervals, and sulfur in diesel fuel at thislevel or lower has been found to have little effect.

Diesel operation and maintenance practice thatincreases smoke emissionsTrucks worldwide have traditionally been loaded tothe maximum possible—often well beyond the de-signed load of the vehicle—in order to increase rev-enue and profit. For example, the principal long dis-tance goods carrier in India is a two-axle, 9-ton truck.But these are often overloaded to 14–20 tons on out-ward-bound trips, although they operate closer tospecifications on the return segment. When trucks areroutinely overloaded, the following effects are typi-cally seen:

� The engine has to work harder to produce thepower needed, resulting in more smoke, espe-cially among older engines.

� Working the vehicle harder tends to increasewear (affecting the engine, tires, brakes, andother vehicle components), but the vehicle is of-ten not serviced any more frequently. As a re-sult, the engine usually operates in a worse stateof repair than a vehicle that is not routinelyoverloaded.

� Because the vehicle requires more power whenoverloaded, there is a tendency to overfuel theengine. Overfueling in turn is a significant causeof higher black smoke emissions.

In industrial countries, professional truck fleetshave learned that they are financially better off in thelong run when they use newer trucks and operatewithin the vehicle specifications. This can reduce theirmaintenance costs and virtually eliminate vehicles

breaking down on the road (a particularly high-costproblem for trucks carrying perishable cargo). Inmany developing countries, however, the underlyingconditions needed to promote improved mainte-nance, including effective enforcement of emissionstandards, do not yet exist.

The impact of overloading on overall emissions iscomplex because the basis for proper comparison isemissions per ton of goods moved. It is not obviousthat one overloaded truck would emit more to movethe same amount of goods as the equivalent numberof properly loaded trucks. Even when vehicles are notoverloaded, however, overfueling on naturally aspi-rated engines is common because it gives more power(and vehicles are often underpowered). Overfuelingleads to higher particulate emissions.

Underpowering and overloading would not beidentified in periodic I/M because heavy vehicles aretypically required to report unloaded. Overfuelingwould also be difficult to catch: truck drivers can re-tune their engines temporarily so as not to beoverfueled and pass emissions tests. For smoke tests,they can also keep injection timing advanced at theexpense of NOx emissions to minimize smoke.

Many automotive repair garages in developingcountries are underequipped, and their workers lackadequate training. The prevalent culture of repairingvehicles when they break down, rather than program-ming preventive maintenance, is both the cause andeffect of the lack of qualified technicians using gooddiagnostic and repair equipment. Rigorously enforc-ing emission standards is the first step toward creat-ing demand for upgrading and expanding serviceand repair facilities.

Malfunctioning and incorrect adjustment (throughtampering or lack of regular vehicle service) of theengine’s air and fuel systems are important causes ofhigh smoke emissions. Dirt, clogging, wear, and incor-rect pressure settings in diesel fuel injectors, togetherwith inadequate repair and calibration of fuel injec-tion pumps and turbochargers, are common causes ofhigh visible exhaust smoke levels. Inadequate anddirty air filters and restrictive exhaust systems will in-crease smoke at high loads when the engine is operat-


ing at near full throttle or accelerating quickly. An en-gine in poor mechanical condition can also consumelubricating oil, thereby emitting higher levels ofsmoke and hydrocarbons. In terms of tampering, in-creasing the fuel delivery of the diesel fuel injectionpump and resetting the engine’s speed governor to ahigher number of revolutions per minute are twocommon practices, both likely to increase smokeemissions.

Another cause of high emissions is the use of infe-rior-quality spare parts. In developing countries, thereare different grades of spare parts, starting with genu-ine original equipment parts with guaranteed quality,followed by varying classes of less expensive partswith lower levels of quality. Most vehicle owners try,within their financial means, to use “closer-to-genu-ine” parts for critical assemblies such as engines andtransmissions, and lower-cost components for non-critical areas such as brakes and electronics.

The engine components that most affect emissionsare probably fuel injection pumps, fuel injectors, andturbochargers. Because of the high cost of these com-ponents, a significant fraction of these parts are re-paired rather than replaced, using components fromalternative sources, some of dubious quality. The useof secondhand injectors from a different engine ver-sion is especially problematic: the injectors may havea different flow rate or release pressure. If the injec-tion pressure is too high or the injector hole size is toosmall, the fuel will penetrate more into the chamber; ifthe fuel hits the chamber wall (in the piston crown),there will be erosion, and the fuel that is cooled bytouching the wall will not be fully combusted. Con-versely, if the injection pressure is too low or the injec-tor hole size is too large, the fuel will not penetrateenough into the chamber. It will have less air to mixwith, adversely affecting both power and emissions.

Impact on the Refining IndustryThe most significant impact of fuel quality specifica-tions on the refining industry in the coming years islikely to concern fuel sulfur reduction. Much of the re-finery investment in North America and Europe thisdecade will be channeled toward reducing sulfur in

gasoline and diesel to ultralow levels. In developingcountries, there are other fuel parameters that can becontrolled to reduce overall vehicle emissions.

The sulfur content of fuels can be lowered by us-ing low-sulfur crude oil or by treating fuels with hy-drogen or both. Some Asian and African crude oilshave relatively little sulfur, producing diesel withclose to 500 ppm sulfur without further treatmentwith hydrogen. But because low sulfur crude oils aremore expensive, there is a trade-off between makingrefinery investments to process high-sulfur crude andpaying more to purchase low-sulfur crude and invest-ing less in refinery upgrade schemes. Worldwide, thesulfur content of crude oil is increasing, requiring ad-ditional refining processing to achieve sulfur reduc-tion. From a refining perspective, reducing sulfur ingasoline tends to be easier and cheaper than reducingsulfur in diesel. Unleaded gasoline typically has amuch lower sulfur content than straight-run diesel.Refineries that rely primarily on reformate for octane,such as those in Pakistan, have very low gasoline sul-fur, although at the cost of having elevated aromaticsand benzene. However, if gasoline contains a lot ofcracked naphtha (which has a high sulfur content), asin Colombia, Mexico, and Peru, it becomes quite ex-pensive to reduce sulfur to ultralow levels even ingasoline.

The level to which a refinery can reduce the sulfurcontent of diesel fuel with relative ease is very muchsituation-specific. For example, because equipment (alow-pressure kerosene hydrotreater and a small dieselhydrotreater) were already in existence, the refineryin Sri Lanka was able to meet the country’s 5,000 ppmsulfur target in 2003 and plans to reduce the sulfurlevel to 3,000 ppm in 2005. Going below 3,000 ppmfor this refinery will require sizable investment and isunlikely to make economic sense unless the refinerysize is substantially increased. In contrast, all of thefour refineries in Pakistan would require significantinvestments to lower diesel sulfur below the currentlevel of about 6,000–7,000 ppm because there are nospare hydrotreating or hydrogen-generation capacities.

The ease of sulfur reduction also depends on fac-tors such as whether or not sulfur “sinks”—products


that do not have to meet lower sulfur specifications—are allowed. In the study of the Chinese refining sec-tor cited in box 2 (Yamaguchi and others 2002), a sig-nificant amount of diesel is used by industry, andindustrial diesel oil served as a sulfur sink for whichthe specifications were not tightened. More impor-tantly, there is a danger that industrial countries willexport difficult-to-upgrade products such as crackednaphtha (high in both sulfur and olefins) to develop-ing countries as they face increasingly stringent fuelspecifications in the latter half of this decade. It is im-portant for developing countries to take steps to en-sure that cheap but dirty blending components fromimports are not increasingly used.

If major reinvestment in large refineries or new in-vestment (again for large refineries) is being under-taken, the incremental costs of going for ultra-low sul-fur fuels as opposed to low sulfur fuels will berelatively small, so that it would make sense to godown to 10 or 15 ppm sulfur in one step. For example,when the Sitra refinery in Bahrain was recently con-sidering revamping to achieve 500 ppm sulfur in die-sel, it realized that it could go to 10 ppm for not muchmore additional cost. Given that it expected to exportmost of its product, going for the higher standard inone step made financial as well as environmentalsense. In examining this case, it is important to bear inmind the scale of the refinery and the project: the re-finery at 250,000 barrels per day (b/d) is large by in-ternational standards, and the diesel output alone isabout 85,000 b/d, larger than many refineries in de-veloping countries. The total cost of the refinery up-grade project is US$0.9 billion, of which about half isfor diesel sulfur reduction.

Scale economics are one of the most significantfactors when considering diesel sulfur reduction to 10ppm. For many existing refineries in developingcountries, severe sulfur reduction of this magnitudecannot be carried out economically, and the two op-tions are (1) refinery closure (a difficult political deci-sion in most circumstances) or (2) significant refineryexpansion. Refinery expansion for a small countrymay not be economic if size expansion means that therefinery has to begin exporting a large fraction of itsproduct slate. However, over the longer term, the

growing demand for fuels in developing countrieswill mean that significant refinery expansion in somecountries is likely, providing an opportunity for fuelquality improvement. For the immediate future, espe-cially where refineries are small and sub-economic forproducing ultralow sulfur fuels, considerable refinerysector restructuring would need to occur in order toproduce these fuels.

There are other alternatives for improving fuelquality that can be pursued without major capital in-vestments. Developing countries with refineries maywish to reexamine their diesel distillation control. Theportion of diesel that boils at high temperature (asmeasured by T90 or T95) can affect particulate emis-sions. High boiling point components can be moredifficult to burn completely, and they also tend to con-tain a greater proportion of PAHs, which have beenshown to increase particulate emissions. High T90 andT95 can also lead to faster and higher deposit buildupsin the engine, particularly in the fuel injectors, causingmalfunctioning and higher smoke and particulateemissions. To avoid greater deposit buildups, morefrequent maintenance of the fuel injection system isneeded. Because of the high diesel-to-gasoline de-mand ratio in developing countries, refiners try to op-timize diesel production by raising T90 or T95. Lower-ing these parameters can affect refinery economicsbut could bring environmental benefits. For dieselwith low cetane numbers, increasing cetane to 45could also lower emissions. Increasing cetane above45 may do little for newer, high-swirl, direct injectionturbo-diesel engines. Increasing the cetane levelabove 45 may help older, naturally aspirated engines.There may be low-cost options for improving fuelquality in the refining process that can be consideredprior to moving to ultralow levels of sulfur.

Different fuel formulations can be used to meetthe same emission targets. For example, sulfur, whichadversely affects the performance of catalytic convert-ers, can be balanced against other fuel parameterssuch as benzene and total aromatics: if the sulfur con-tent is extremely low, catalytic converter efficiencywill increase, converting more benzene and other aro-matics in the engine-out exhaust, thereby reducingemissions. In order to reduce the cost of emission re-


duction, refiners in the United States have been giventhe flexibility to select an optimal combination of fuelparameters to meet the same emission standards.While monitoring fuel quality may be more compli-cated in such a system, this flexibility has enabledU.S. refiners to reduce emissions at lower costs than ifgasoline formulation were precisely specified. More-over, it is apparent that different components are asso-

ciated with different pollutants and pollution prob-lems. What is important in one country may be lessimportant in another. Domestic problems and vulner-abilities should be reflected in national gasoline stan-dards. Whenever feasible, refiners should be allowedto meet specific emission standards rather than haveprecise fuel composition specified, which can increasethe cost of regulation.

103

Trends in Vehicular Emission Standardsand Fuel Specifications in the United States

This annex discusses vehicular emission standards inthe United States. The U.S. standards have a signifi-cant impact on Latin America and the Caribbean and,together with EU standards, give a good indication ofthe direction in which future standards are likely todevelop in the rest of the world. A major auto and oilindustry study undertaken in the United States istreated at some length, as the findings provide valu-able data on the complex relationships between ve-hicular emissions, gasoline quality, alternative fueluse, and vehicle technology. In interpreting the num-bers presented in this annex, it is important to bear inmind that the driving cycle used to generate emissiondata has a significant impact on the emission levels.Reference fuels used are also different, although theirimpact would normally not be as large as that fromdifferences in the driving cycle. For these reasons, theemission standards elsewhere, such as the EU or Ja-pan, should not be compared directly with those pre-sented here on an absolute numerical basis.

The U.S. Congress passed the first major Clean AirAct (CAA) in 1970, requiring a 90 percent reduction inemissions from new automobiles by 1975 and estab-lishing the U.S. EPA and assigning it broad responsi-bility for regulating motor vehicle pollution. The re-quired standards were, however, not met until 1981.Congress amended the Clean Air Act in 1990 and re-quired further reductions in hydrocarbon, CO, NOx,and particulate emissions. The amendments also in-troduced more stringent emission testing procedures,expanded I/M programs, new vehicle technologiesand clean-fuels programs, and transportation man-agement programs (U.S. EPA undated a).

The emission standards for heavy-duty vehicleswere introduced in 1970 (U.S. EPA undated b). The

first particulate emission standards were introducedfor 1988 model year (MY) (autumn 1987) and tight-ened progressively for 1991 and 1994 MY.

The U.S. reformulated gasoline (RFG) programcame into force in 1995. As of the early 2000s, RFGconstituted about 35 percent of the U.S. gasoline sup-ply, the rest being “conventional” gasoline with muchless stringent specifications. The RFG program givesmore flexibility than perhaps any other programworldwide, allowing refiners to seek the least-cost ap-proach to meeting vehicular emission standards. TheU.S. auto and oil industry study generated a largeamount of data correlating vehicular emissions, gaso-line quality, vehicle technology, and air quality. Thevehicular emission and fuel quality standards will beconsiderably tightened in the United States with theintroduction of the so-called Tier 2 emission stan-dards, which come into effect beginning in 2004 forgasoline vehicles and in 2006 for diesel vehicles.

Clean Air Act Amendments of 1990The CAA Amendments of 1990 required significantchanges in the U.S. refining industry. The amend-ments defined two categories of regulated gasoline—oxygenated gasoline (OxyFuel) and RFG. OxyFuel,which is gasoline with an oxygen content of 2.7 per-cent by weight (wt%)—corresponding to 15 percentby volume (vol%) methyl tertiary-butyl ether or 7.3vol% ethanol—is specified for CO nonattainment ar-eas1 during the winter months, when CO emissions

1Nonattainment areas are those areas that are not in compli-ance with national air quality standards. Separate regulations ap-ply to CO and ozone nonattainment areas. Noncompliance withCO standards is primarily a winter problem. Noncompliance withozone standards is primarily a summer problem in the UnitedStates.

A N N E X2


are high. “RFG” refers to a more extensive change ingasoline properties that reduces VOC emissions andtoxic emissions. RFG is required in the areas in theUnited States that have the most serious problemswith ozone pollution. RFG requires 2 wt% oxygenthroughout the year, corresponding to 11 vol% MTBEor 5.4 vol% ethanol.

The CAA Amendments require vehicle emissionreductions in two phases. For Phase I, beginning in1995, the law specified a minimum of 15 percent re-duction in VOC emissions during the high-ozone sea-son and a minimum of 15 percent reduction in toxinemissions during the entire year. The U.S. EPA esti-mates that the impact of Phase I of the RFG programwas to reduce annual emissions by 17, 2, and 17 per-cent for VOCs, NOx, and air toxics, respectively, since1995. Under Phase II, which began in 2000, the pro-gram will cut VOC, NOx, and air toxics emissions by27, 7, and 22 percent, respectively, compared with1995 (U.S. EPA 1999c). The baseline gasoline in 1990and fuel specifications to meet these reductions aregiven in table A2.1–table A2.3, and federal diesel stan-dards in table A2.4.

As table A2.2 and table A2.3 show, Phase I wasfurther divided into two periods, using the SimpleModel and the Complex Model, respectively. TheSimple Model established three different sets of equa-tions to calculate toxics reduction from the baselinegasoline. The Complex Model was more detailed,

U.S. Industry Average Baseline Gasoline, 1990

Gasolineparameter Unit Summer Winter

RVP psi 8.7 11.5Benzene vol% 1.53 1.64Total aromatics vol% 32.0 26.4Olefins vol% 9.2 11.9Sulfur wt ppm 339 338Oxygen wt% 0.0 0.0

psi Pounds per square inch.

Source: UOP 1994.

T A B L EA 2. 1

Simple Model, 1 January 1995–31 December 1997

RFGGasoline parameter Unit Per gallon Averaging Conventional gasoline

Summer RVP, maximum psi 7.2a / 8.1b 7.1a / 8.0b —Benzene, maximum vol% 1.0 0.95 —Oxygen, minimum wt% 2.0 2.1 —Air toxics emission reduction % 15.0 16.5 —Exhaust benzene emissions — Same as refinery’s 1990 gasolineSulfur, olefins and T

90Capped at refinery’s 1990 levels Capped at 1.25 × refinery’s 1990 levels

psi Pounds per square inch; — not applicable.

aVOC control region 1, southern half of the United States.bVOC control region 2, northern half of the United States.

Source: UOP 1994.

T A B L EA 2. 2

consisting of different systems of equations to calcu-late VOC, NOx, and toxics emissions. A different set ofComplex Model equations have been established forPhase II. There are minimal numerical requirementson fuel composition in both cases.

The above illustrates one prominent feature of theU.S. fuel specifications, in sharp contrast to the rest ofthe world: they are performance based. The specifica-tions rely on empirically derived models to identify arange of fuel compositions that will achieve emissiontargets. They require an extensive database of emis-sion levels as a function of fuel composition and ve-hicle characteristics. The empirical relationships mayhave to be updated from time to time as the vehiclefleet characteristics evolve over time. Monitoring is

ANNEX 2. TRENDS IN VEHICULAR EMISSION STANDARDS AND FUEL SPECIFICATIONS IN THE UNITED STATES 105

Complex Model, 1 January 1998–31 December 1999

RFGGasoline parameter Unit Per gallon Averaging Conventional gasoline

Benzene, maximum vol% 1.0 0.95 —Oxygen, minimum wt% 2.0 2.1 —VOC reduction % 35.1a/15.6b 36.6a/17.1b —NO

x reduction % 0.0 1.5 —

Air toxics emission reduction % 15.0 16.5 —Exhaust toxics emissions — Same as refinery’s 1990 gasoline

psi Pounds per square inch; — not applicable.

aVOC control region 1, southern half of the United States.bVOC control region 2, northern half of the United States.

Source: UOP 1994.

T A B L EA 2. 3

more complicated, because the compositional analysisof the fuel must be checked against empirical equa-tions. These standards may hence be more expensiveto implement and enforce than composition-basedstandards, which are currently in force in developingcountries.

Once the mathematical models are set up, how-ever, these specifications offer far greater flexibility torefiners, enabling them to select the most economicway to meet emission targets, which, in turn, depend

on the air quality of a given region. This approach hasthe potential of providing a significant cost advantageto the refining sector in the United States, making itless expensive to meet clean air targets than if therehad been uniform, nationwide, composition-basedfuel specifications.

Another example of the flexibility offered by per-formance-based standards is the California dieselregulations. California has been a leader in emissioncontrol legislation and has generally adopted limitsmore severe than the federal CAA limits that apply tothe rest of the United States. With respect to diesel, theCalifornia Air Resources Board (CARB) adopted adiesel fuel specification of 500 wt ppm sulfur and 10vol% aromatics from October 1993. It is important torecognize that there are different ways of certifyingdiesel in California. Reducing aromatics to 10 percentwould be very costly, and an alternative approachused in California is explained in box 3. According toCARB, most refiners are not complying with the de-fault 10 percent aromatics limit in the CARB dieselrule, but instead use certified alternative formulasthat usually include achieving a high cetane numberby means of cetane improvers while reducing sulfur(CARB 2000a). In designing performance-based fuelspecifications, it is important that the performancemeasures be representative of real vehicle and fueluse.

Federal Diesel Standards

Low sulfur Low sulfurGasoline parameter Unit No. 1-D No. 2-D

Cetane number, minimum 40 40T

90, minimum °C 288 338

Kinematic viscosity at 40°C cSt 1.3–2.4 1.9–4.1Sulfur, maximum wt ppm 500 500Cetane index, minimuma 40 40Aromaticity, maximuma vol% 35 35Flash point, minimum °C 38 52Ash, maximum wt% 0.01 0.01Ramsbottom carbon on wt% 0.15 0.35 10% residue, maximum

cSt Centistokes.

aOne of the two properties must be met.

Source: ASTM 1997.

T A B L EA 2. 4


Because cetane improvement additives are effec-tive in increasing the cetane number but not cetane in-dex, diesel specifications should give the option of us-ing either the cetane number or index, but notmandate the minimum for the cetane index only.

Up until now, MTBE has been widely used ingasoline in the United States, in part to meet the mini-mum oxygen content requirement. In March 1999, thegovernor of California announced that Californiawould begin immediate phase-out of MTBE fromgasoline, with complete elimination to be achieved nolater than 31 December 2002 (later postponed to2003). The principal reasons cited were MTBE con-tamination of lakes, particularly in recreational areas,2

as well as fears that MTBE may be making its way insignificant amounts into the state’s groundwater sup-

Diesel Certification in CaliforniaB O X 3

The importance of understanding how different fuel compositions can

meet the same vehicle emission standards can be seen in the Califor-

nia program regarding NOx emissions and diesel fuel composition. Cali-

fornia issued specifications for diesel fuel that set a maximally permis-

sible level of 500 wt ppm for sulfur and 10 percent for aromatics. In

addition to the fuel specification, California also offers an alternative

method for certifying a diesel fuel—namely, that a fuel produce the

same NOx reduction when compared with a California reference fuel

run in a Detroit Diesel Series 60 engine. The California reference fuel

controlled many more parameters than the specification. These param-

eters are shown below. Also shown are four alternative diesel fuel for-

mulations certified by CARB that are equivalent to the California refer-

ence in NOx reduction. The four alternative diesel formulations are

much less costly to produce (because they require less severe

hydrotreating) than the California reference.

Diesel Formulations Certified by CARB

California Chevron ARCODiesel property reference G2 D4781 D-25 D-26

Maximum vol% aromatics 10.0 15.1 19.0 21.7 24.7Maximum wt% PAHs 1.4 3.6 2.2 4.6 4.0Maximum wt ppm sulfur 500 202 54 33 42Maximum wt ppm nitrogen 10 341 484 20 40Minimum cetane number a 48.0 54.8 58.0 55.2 56.2

aWith cetane-improvement additives.

Source: CARB undated.

plies.3 As of June 2002, 15 other states had announcedtheir intention to ban MTBE eventually. As shown inannex 3, the EU has taken a different position onMTBE.

Air Quality Improvement ResearchProgramThe U.S. Air Quality Improvement Research Program(AQIRP)—conducted by three U.S. auto companiesand 14 oil companies at a cost of US$40 million be-tween 1989 and 1995—was set up to provide data tohelp legislators and regulators achieve the nation’sclean air goals through a research program consistingof (1) extensive vehicle emission measurements, test-ing vehicles as old as MY 1983 as well as “advanced”prototype vehicles; (2) air quality modeling studies topredict the effects of the measured emissions on2The dispute over MTBE has become particularly acute in rec-

reational areas where large numbers of personal watercraft suchas Jet Skis and small boats are common, many of them poweredby two-stroke engines. According to some estimates, up to 30 per-cent of the fuel in the Jet Skis’ two-stroke engines is released un-burned into the water.

3One study by researchers at the Lawrence Livermore Na-tional Laboratory concluded that more than 10,000 sites havebeen contaminated by MTBE since 1992.


ozone formation; and (3) economic analysis of someof the fuel/vehicle systems. The program focused ongasoline and alternative fuels but did not cover diesel.It was the largest and most comprehensive researchprogram of this nature ever conducted in the UnitedStates. Some of the findings are summarized in theparagraphs that follow.

OzoneReducing aromatics (note: those with two or morealkyl branches are ozone precursors) in gasoline from45 percent to 20 percent had no statistically significantimpact on predicted ozone. For vehicles not equippedwith catalytic converters, increasing aromatics ingasoline increased NOx emissions. NOx is an ozoneprecursor.

Fuel composition changes that reduced the pre-dicted ozone contributions of light-duty vehicles wereT90 and T50 (the temperature at which 50 percent ofgasoline evaporates) reduction, olefin reduction, sul-fur reduction (for catalyst-equipped cars), and volatil-ity (RVP) reduction.

Reducing gasoline olefins from 20 percent to 5percent increased exhaust hydrocarbon emissions andreduced NOx emissions, reduced the photochemicalreactivity of exhaust and evaporative emissions, andled to a marked decrease in predicted ozone.

Decreasing RVP from 9 pounds per square inch(psi) (62 kilopascals, kPa) to 8 psi (55 kPa) reducedevaporative emissions as well as exhaust HC and CO,and led to a marked decrease in predicted ozone.

Gross emittersHigh-emitting, poorly maintained vehicles on theroad contributed about 80 percent of total vehicularemissions but represented only about 20 percent ofthe vehicle population. Identification and repair ofthese vehicles can result in substantial emission re-ductions. Among gross emitters, there are tailpipegross emitters, evaporative gross emitters, and “drip-pers” leaking fuel.

Air toxicsOf the four CAA-designated toxic air pollutants mea-sured—benzene, 1,3-butadiene, formaldehyde, and

acetaldehyde—benzene (a carcinogen) had the largestconcentrations in the emissions. Decreasing fuel ben-zene or the total aromatic content reduced benzeneemissions.

OxygenatesAdding oxygenates to gasoline reduced exhaust HCand CO in the 1989 and earlier vehicle models, andraised NOx with low aromatic fuels. The 1993 andlater model year vehicles did not show any emissionchange. This is to be expected because later modelyear cars are equipped with oxygen sensors (whichlet the vehicle adapt to the fuel type to maintain a tar-get oxygen-to-fuel ratio).

SulfurDecreasing sulfur generally reduced exhaust toxics,hydrocarbon, CO, and NOx for cars equipped withthree-way catalysts.

In the above findings, of particular interest is thesurprising result that reducing the amount of aromat-ics more than two-fold in gasoline had no impact onpredicted ozone. Since aromatics are an importantsource of octane in gasoline as well as hydrogen(which is needed to reduce sulfur in fuels) at refiner-ies, dramatic aromatics reductions would have a con-siderable adverse impact on refinery economics whilenot having much benefit in the way of air quality im-provement once the vehicle fleet is equipped withcatalytic converters. This implies that once the major-ity of gasoline-fueled vehicles are equipped with cata-lytic converters, the only significant adverse healthimpact of aromatics will be benzene emissions. How-ever, for benzene control, it is typically more cost-ef-fective to control benzene in gasoline than total aro-matics.

Tier 1 and Tier 2 Emission StandardsThe so-called Tier 1 vehicle emission standards forlight-duty vehicles in the United States were intro-duced progressively from 1994. Starting in 1996, ve-hicles have had to be certified up to 100,000 miles(160,000 km) or to the higher of “useful-life” limits.The durability of the emission-control device must be


demonstrated over this distance, with allowed dete-rioration factors. The heavy-duty truck regulations for1987 and later require compliance over longer peri-ods, representative of the useful life of the vehicle.The exhaust emission standards for light-duty ve-hicles are shown in table A2.5.

The standards for compression ignition engine ve-hicles are shown in table A2.6. The manufacturer war-ranty period is five years or 160,000 km, whichever

Tier 1 U.S. Federal Exhaust Emission Standards for Light-Duty Vehicles, Federal Test Procedure, Cold CO (g/km)

Vehicle type NMHC CO NOx

PM NMHC CO NOx

PM

5 years or 80,000 km 10 years or 160,000 km

LDV 0.16 2.13 0.25 0.05 0.19 2.63 0.38 0.06LDT1 0.16 2.13 0.25 0.05 0.19 2.63 0.38 0.06LDT2 0.20 2.75 0.44 0.05 0.25 3.44 0.61 0.06

5 years or 80,000 km 11 years or 192,000 km

LDT3 0.20 2.75 0.44 — 0.29 4.00 0.61 0.06LDT4 0.24 3.13 0.69 — 0.35 4.56 0.96 0.08

NMHC Nonmethane hydrocarbons; LDV light-duty vehicle, a passenger car or car derivative capable of seating 12 passengers or less; LDT1light-duty truck 1, any light light-duty truck up through 3,750 pounds (1,705 kg) loaded vehicle weight; LDT2 light-duty truck 2, any light light-dutytruck greater than 3,750 pounds loaded vehicle weight; LDT3 light-duty truck 3, any heavy light-duty truck up through 5,750 pounds (2,611 kg)adjusted loaded vehicle weight; LDT4 light-duty truck 4, any heavy light-duty truck greater than 5,750 pounds adjusted loaded vehicle weight; — not applicable.

Source: U.S. EPA 2000b.

T A B L EA 2. 5

comes first. The useful life of the engine, over whichcompliance with emission standards has to be dem-onstrated, is as follows for the 1987–2003 model years:

� Light heavy-duty diesel engines: 8 years or176,000 km (whichever occurs first)

� Medium heavy-duty diesel engines: 8 years or296,000 km

� Heavy heavy-duty diesel engines: 8 years or464,000 km.

U.S. Federal Heavy-Duty Exhaust Emission Standards Compression Ignition and Urban Buses (g/kWh)

Year CO HC NMHC+NOx

NOx

PM Smokea

1990 20.8 1.7 — 8.0 0.80 20/15/501991–1993 20.8 1.7 — 6.7 0.34 / 0.13b 20/15/501994–1997 20.8 1.7 — 6.7 0.13 / 0.09c / 0.07d 20/15/501998+ 20.8 1.7 — 5.4 0.013 / 0.07d 20/15/502004+ 20.8 — 3.2 or 3.4 with a limit — 0.13 / 0.07d 20/15/50

of 0.7 on NMHC2007 20.8 0.19 for NMHCe — 0.27e 0.013 20/15/50

NMCH Nonmethane hydrocarbons; — not applicable.

aData are in percentages; these apply to smoke opacity at acceleration/lug/peak modes.bStandards for urban buses for 1993.cStandard for urban buses for 1994-95.dStandard for urban buses from 1996 and later.ePhased in together between 2007 and 2010.

Sources: U.S. EPA 1997a and 2000d.

T A B L EA 2. 6


The useful life requirements were later increasedto 10 years, with no change to the above mileagenumbers, for the urban bus PM standard (1994+) andfor the NOx standard (1998+). Beginning with the2004 model year, EPA revised useful engine life as fol-lows:

� Light heavy-duty diesel engines: 10 years or176,000 km

� Medium heavy-duty diesel engines: 10 years or296,000 km

� Heavy heavy-duty diesel engines: 10 years or696,000 km or 22,000 hours.

The emission levels of new vehicles are measuredin a certified driving cycle using reference fuels ofspecific compositions covering limited ranges forvarious fuel parameters. A test driving cycle calledFTP75 (Federal Test Procedure 75) has been used inthe United States since 1975. As required by the CAAAmendments, the U.S. EPA re-evaluated typical driv-ing patterns and found that the FTP test cycle doesnot cover about 15 percent of driving conditions. As aresult EPA issued a Final Rule in August 1996 settingout modifications. The main element of this rule is aSupplemental Federal Test Procedure (SFTP), cover-ing the driving patterns not included in FTP75. Thiswas to address a potential concern that becameknown as “cycle beating.” Cycle beating refers to sce-narios whereby vehicle manufacturers design vehiclesto meet the emission standards in FTP75, but whereemission levels increase significantly in other drivingcycles. SFTP includes two new driving cycles, onerepresenting aggressive and microtransient driving,and another representing driving immediately follow-ing vehicle startup.

In order to meet the Tier 1 emission standards forparticulate emissions from diesel-fueled vehicles inthe United States, a sulfur limit of 500 wt ppm wasimposed from October 1993. Up until that point, par-ticulate emissions were controlled by continuous im-provement in vehicle technology without loweringdiesel sulfur below 2,500 wt ppm (0.25 wt%). This isillustrated in annex 1, figure A1.1.

Beginning with the 1994 MY (autumn 1993), light-duty vehicles and light-duty trucks have been re-

quired to be equipped with on-board diagnostic(OBD) systems. OBD systems monitor emission con-trol components for any malfunction or deteriorationthat cause emission limits to be exceeded, and alertthe driver of the need for repair via a dashboard lightwhen the diagnostic system has detected a problem.EPA made changes to the federal OBD requirementsstarting in the 1999 MY (autumn 1998). The modifica-tions include harmonization of the emission levelsabove which a component is considered malfunction-ing with California’s OBD Generation Two (OBD II)requirements. OBD systems are generally seen as acomplement to traditional I/M programs rather thana substitute. By January 2001, all areas with basic andenhanced I/M programs were required to implementOBD checks as a routine part of I/M programs. Fail-ure of the OBD test would require mandatory repair.In fact, in Maryland, for example, the OBD II checklight is a substitute for a chassis dynamometer emis-sions test.

In the future, Tier 2 vehicle emission standardswill aggressively pursue significant emission reduc-tions. Tier 2 is a comprehensive national control pro-gram that regulates the vehicle and its fuel as a singlesystem. The new tailpipe emission standards for pas-senger vehicles will reduce NOx emissions by 77 per-cent from cars and up to 95 percent from SUVs andtrucks. The standards for SUVs and trucks will bebrought in line with those of other cars for the firsttime, and the same standards will be applied to gaso-line, diesel, methanol, ethanol, natural gas, and LPGfuels. The exhaust emission standards for these light-duty vehicles are given in table A2.7 (only permanent,and not temporary, certification bins are shown) andwill be phased in between 2004 and 2009, as shown intable A2.8. The exhaust emission standards are struc-tured into eight certification levels called “certificationbins.” Vehicle manufacturers can choose any of theeight bins, but must meet the average NOx standardfor the entire fleet of 0.04 g/km (0.07 grams per mile).The vehicle “full useful life” has been extended to192,000 km.

The corresponding schedule for gasoline sulfurstandards is given in table A2.9. Sulfur in gasoline isrequired to be reduced to a mandatory average of 30


wt ppm. “Small refiners” (with an average crude ca-pacity of less than 155,000 b/d and employing fewerthan 1,500 people) are eligible for hardship provisionsthat permit an annual average sulfur limit of up to300 wt ppm until 31 December 2007. The cost to con-sumers is estimated by EPA to be less than US$100 forcars, $200 for light-duty trucks, and less than $0.02per gallon (0.5 U.S. cents per liter) for gasoline sulfurreduction, with the overall cost to industry on the or-der of $5.3 billion against health and environmentalbenefits of about $25 billion (U.S. EPA 1999d).

For gasoline sulfur reduction, U.S. EPA has basedits incremental cost calculations on two newhydrodesulfurization technologies, Mobil’s Octgainand CD Tech’s CDHydro/CDHDS, neither of whichhad been commercially proven at the time the incre-mental costs were computed. For the U.S. refining in-dustry overall, the difference in capital investment be-tween conventional technologies and emergingtechnologies may be dramatic: estimates have variedby a factor of two or more. Separate refinery model-ing exercises by the EPA, the auto industry, and the oilindustry confirm that newer desulfurization technolo-gies are nearly 50 percent less costly than older tech-nology. EPA concluded the cost of achieving 30 ppmsulfur would initially be 1.95 (1997) U.S. cents per gal-lon (0.5 U.S. cents per liter), while the newer technol-ogy would achieve the same reduction at a cost of

Phase-in Percentages for Tier 2 EmissionStandards for Light-Duty Vehicles, Light-DutyTrucks, and Medium-Duty Passenger Vehicles

Percentage of vehiclesthat must meet Tier 2

Model year requirements

Light-duty and light light-duty trucks2004 252005 502006 752007 and subsequent 100

Heavy light-duty trucks and medium -dutypassenger vehicles

2008 502009 100

Source: U.S. EPA 2000c.

T A B L EA 2. 8

Tier 2 FTP Exhaust Emission Standards for Light-Duty Vehicles, Light-Duty Trucks, and Medium-Duty PassengerVehicles, Permanent (g/km)

80,000 km 192,000 kmBin NMHC CO NO

xPM HCHO NMHC CO NO

xPM HCHO

8 0.063 2.1 0.088 — 0.0094 0.078 2.6 0.125 0.0125 0.01137 0.047 2.1 0.069 — 0.0094 0.056 2.6 0.094 0.0125 0.01136 0.047 2.1 0.050 — 0.0094 0.056 2.6 0.063 0.0063 0.01135 0.047 2.1 0.031 — 0.0094 0.056 2.6 0.044 0.0063 0.01134 — — — — — 0.044 1.3 0.025 0.0063 0.00693 — — — — — 0.034 1.3 0.019 0.0063 0.00692 — — — — — 0.006 1.3 0.013 0.0063 0.00251 — — — — — 0.000 0.0 0.000 0.0000 0.0000

NMHC Nonmethane hydrocarbons; HCHO formaldehyde; — not applicable.


T A B L EA 2. 7

Gasoline Sulfur Limits in the United States

Averaging period beginningJan. Jan. Jan. 2006 and2004 2005 subsequent

Refinery or importer — 30 wt ppm 30 wt ppmaverageCorporate pool 120 wt ppm 90 wt ppm —averagePer-gallon limit 300 wt ppm 300 wt ppm 80 wt ppm

— Not applicable.


T A B L EA 2. 9


1.26 U.S. cents per gallon (0.3 U.S. cents per liter) (U.S.EPA 1999e).

In July 2000, U.S. EPA issued a final rule for thefirst phase of the program on heavy-duty trucks andbuses, taking effect beginning in 2004 (U.S. EPA2000a). The emission standards for heavy-duty dieselvehicles—a combined standard for NOx and HC of3.2 g/kWh—represent a reduction of more than 40percent in emissions of NOx as well as HC. The rulerequires OBD systems for engines between 8,500 and14,000 pounds (3,864 and 6,364 kg) to be phased in,beginning in 2005. Heavy-duty gasoline vehicles lessthan 6,364 kg are subject to emission standards andtesting similar to the current program for light-dutyvehicles and light-duty trucks. As a result of a legalsettlement reached between six engine manufacturersand the government, the date for meeting the 2004emission standards was advanced 15 months to Octo-ber 2002 (see chapter 3, Vehicle Technology, Vehicleemissions-control technologies). The emission stan-dards are shown in table A2.10. EPA estimated that anaverage projected long-term incremental cost of thisprogram would be less than US$400 per vehicle forheavy-duty diesel engines and less than $300 per ve-hicle for heavy-duty gasoline engines.

In December 2000, U.S. EPA also signed emissionstandards for MY 2007 and later heavy-duty highwayengines (U.S. EPA 2000d). The regulation introducesnew, very stringent exhaust emission standards: 0.013g PM per kWh, 0.27 g NOx per kWh, and 0.19 g non-methane hydrocarbons (NMHC) per kWh. This is thesecond phase of the control program for heavy-duty

vehicles and will achieve emission reductions of up-wards of 90 percent over levels achieved by the Phase1 reductions described in the preceding paragraph.The PM emission standard will take full effect for die-sels in the 2007 MY. The NOx and NMHC standardswill be phased in for diesel engines between 2007 and2010. The phase-in would be on a percent-of-sales ba-sis: 50 percent from 2007 to 2009 and 100 percent in2010. Gasoline engines will be subject to these stan-dards based on a phase-in requiring 50 percent com-pliance in 2008 and 100 percent compliance in 2009.Emission certification requirements also include theSET (supplementary emission test),4 with limits equalto the FTP standards, and NTE (not to exceed)5 limitsof 1.5 × FTP standards.

Diesel sulfur is limited to 15 wt ppm, down fromthe previous limit of 500 wt ppm. Refiners will be re-quired to start producing the 15 wt ppm sulfur fuelbeginning 1 June 2006. At the terminal level, highwaydiesel fuel sold as low-sulfur fuel must meet the 15 wtppm sulfur standard as of 15 July 2006. For retail sta-tions and wholesale purchasers, highway diesel fuelsold as low sulfur fuel must meet the 15 ppm sulfurstandard by 1 September 2006. This limit is based onEPA’s assessment of how sulfur-intolerant advancedaftertreatment technologies will be. EPA estimates thecost of the program to be about US$1,200 to US$1,900per new vehicle, and the incremental production anddistribution cost of lowering sulfur from the currentlimit of 500 wt ppm to 15 wt ppm to be approxi-mately US$0.045–0.05 per gallon (1.2–1.3 U.S. centsper liter).

In response to these new rulings and proposals,the automotive industry is confident of being able to

Heavy-Duty Gasoline Exhaust EmissionStandards for 2004 and Later Model Year

GVWR Old NewFrom to HC NO

xHC NO

xHC+NO

x

kg kg g/kWh g/kWh g/km g/km g/kWh3,864 4,545 1.5 5.4 0.18 0.56 —4,545 6,364 1.5 5.4 0.21 0.63 —6,364 — 2.5 5.4 — — 1.3

GVWR Gross vehicle weight rating; — not applicable.

Source: U.S. EPA 2000a.

T A B L EA 2. 1 0

4The SET is a 13-mode steady-state test that was introduced tohelp ensure that heavy-duty engine emissions are controlled dur-ing steady-state type driving, such as a line-haul truck operatingon a freeway. The test is based on the EU 13-mode European sta-tionary cycle (ESC) schedule (commonly referred to as the “EuroIII” cycle in the United States).

5The NTE approach establishes an area under the torquecurve of an engine where emissions must not exceed a specifiedvalue of any of the regulated pollutants. The NTE requirementwould apply under any engine operation conditions that couldreasonably be expected to be seen by that engine in normal vehicleoperation and use, as well as a wide range of real ambient condi-tions.


meet Tier 2 standards for cars, but is less certain aboutsport utility vehicles (SUVs). As for diesel vehicles,particulate control technology is considered feasiblealthough particulate filters have not yet been demon-strated on low-power operation vehicles in cold cli-mates. NOx emission control is not yet proven andthere is no general consensus that the 2007 NOx emis-sion limits can be met and be durable for the enginelife. What the automotive industry is especially con-cerned about are the supplementary requirements,and especially the NTE requirements, which are pro-posed to take effect starting in the 2007 model year.This illustrates the interdependence between testdriving cycles and emission levels and the importanceof not concentrating on establishing emission levelsonly.

As for sulfur levels, a significant consideration inthe United States is cross-contamination of low sulfurfuels, particularly diesel, in the distribution system byhigher sulfur fuels. If chances of cross-contaminationare not negligible, having to ensure a maximum of 15

wt ppm sulfur in diesel at the pump would meanmuch lower sulfur levels at the refinery gate. For ex-ample, pipeline operators expect that refiners willhave to provide diesel fuel with sulfur levels as low as7 wt ppm in order to compensate for possible con-tamination from higher sulfur products in the system.Even at 7 wt ppm, industry representatives believethe likelihood of contamination during the delivery ofthe fuel through the distribution system is extremelyhigh. Once contamination occurs, the product cannotbe sold as automotive diesel fuel, thereby diminishingthe fuel supply (U.S. GAO 2004).

Another concern is the contribution of sulfur,phosphorus, and ash found in lubricants whichwould cause legislation compliance problems if thelevels found in today’s lubricants are maintained. Vis-cosity modifiers normally contain sulfur, and anti-oxi-dant additives can contain phosphorus. Straightfor-ward reductions in sulfur and phosphorus can harmpiston cleanliness and shorten the life of the lubricant(Bunting 2003).

113

Trends in Vehicular Emission Standards andFuel Specifications in the European Union

This annex discusses vehicle emission standards inthe European Union (EU). These standards have asignificant impact on Eastern Europe, the former So-viet Union Republics, and Asia. The first set of direc-tives for controlling emissions from motor vehicleswas issued in Council Directive 70/220/EEC of 20March 1970. This directive has been amended numer-ous times since, increasingly tightening standards.

A major auto and oil industry study undertaken intwo phases, examining both gasoline and diesel (al-though the U.S. auto and oil study did not examinediesel vehicles), is treated at some length in this an-nex. The findings provide information on the complexrelationships between vehicular emissions, fuel qual-ity, and vehicle technology. There are useful lessonsfor developing countries from these studies, dis-cussed below.

European Auto-Oil ProgrammeAt the end of 1992, the European Commission invitedthe European auto and oil industries to participate ina technical program to carry out an objective assess-ment of cost-effectiveness of different measures for re-ducing emissions from the road transport sector. Thiseffort became known as the “Auto-Oil Programme.”Its scope included studying the future developmentof emissions from the European vehicle fleet, model-ing air quality under different policy scenarios, exam-ining vehicle–fuel quality interactions (the EuropeanProgramme on Emissions, Fuels and Engine Tech-nologies), and estimating cost-effectiveness of differ-ent mitigation options.

The first Auto-Oil Programme was concluded in1996 and was instrumental in shaping subsequentemission and fuel quality standards. Auto-Oil I con-

firmed that the relationships found among fuel prop-erties, engine technologies, and exhaust emissions arecomplex. Changes in a given fuel property may lowerthe emissions of one pollutant but increase those ofanother. (For example, decreasing aromatics in gaso-line lowered CO and HC emissions but increased NOx

emissions for catalyst-equipped cars.) In some cases,engines in different vehicle categories, such as heavy-duty and light-duty vehicles, had opposite responsesto changes in fuel properties. (For example, reducingPAHs in diesel reduced HC emissions in heavy-dutyengines but increased HC, CO, and benzene emis-sions in light-duty vehicles.)

With respect to both gasoline and diesel vehicles,individual vehicles and engines showed a wide rangeof response to the fuel properties investigated. In thecase of gasoline vehicles, some vehicles that showedlow fuel sensitivity for CO and HC emissions showedhigh sensitivity for NOx and vice versa. In the case ofdiesel vehicles, the impact of the vehicle/engine seton emissions was larger than that of the matrix of fuelproperties except for NOx emission on heavy-duty en-gines. These findings underscore the importance oftargeting the vehicle hardware.

European Programme on Emissions, Fuels andEngine TechnologiesThe European Programme on Emissions, Fuels andEngine Technologies (EPEFE) was designed to en-hance the understanding of the relationships betweenfuel properties and engine technologies and to quan-tify the reduction in vehicular emissions that can beachieved by combining advanced fuels with the ve-hicle/engine technologies (ACEA and Europia 1995).The EPEFE marks an unprecedented cooperation be-tween the European motor industry (represented by

A N N E X3


the Association des Constructeurs Européensd’Automobiles [ACEA]) and the European oil indus-try (represented by EUROPIA, the European govern-ment affairs organization of the oil refining and mar-keting industry in the EU and the European EconomicArea). Unlike the U.S. Air Quality Improvement Re-search Program (AQIRP), EPEFE examined diesel inaddition to gasoline.

The results concerning diesel fuel and vehicle in-teractions from EPEFE should be interpreted withcaution because the ranges of fuel parameters exam-ined often fell outside of the ranges found in manydeveloping countries and typically represent much“cleaner” diesel. Decreasing density from 857 kilo-grams per cubic meter (kg/m³) to 829 kg/m³ de-creased PM emissions in light-duty vehicles and NOx

emissions in heavy-duty engines, but increased NOx

emissions in light-duty vehicles and HC and COemissions in heavy-duty engines. Decreasing densityhad no significant effect on PM emissions fromheavy-duty engines. EPEFE examined the impact ofvarying PAHs on vehicular emissions and found thatdecreasing PAHs from 8 percent to 1 percent de-creased both particulate and NOx emissions fromlight-duty and heavy-duty diesel vehicles. LoweringT95 from 370°C to 325°C had mixed effects, increasingHC and CO emissions from heavy-duty engines andNOx from light-duty vehicles, but decreasing particu-late emissions from light-duty vehicles and NOx fromheavy-duty engines. Increasing cetane number from50 to 58 decreased NOx emissions from heavy-dutyengines only, increased particulate emissions in light-duty vehicles with no significant effect on heavy-dutyvehicles, and decreased HC and CO emissions fromboth light-duty vehicles and heavy-duty engines.

Examination of vehicle and fuel technologyinteractions under Auto-Oil IIThe Auto-Oil II Programme ran from 1997 to 2000 asa follow-up to Auto-Oil I (European Commission2000a). The two principal aims were to (1) assess fu-ture air quality and establish a consistent frameworkwithin which different policy options to reduce emis-sions can be assessed using the principles of cost-ef-fectiveness, sound science, and transparency; and (2)

provide a foundation for the transition toward longer-term air quality studies covering all emission sources.

The air quality modeling in Auto-Oil II suggestedthat meeting the particulate air quality standardposed the greatest challenge during the period 2000–2010. Special attention was paid to measures thatcould reduce PM emissions from diesel vehicles. Theprogram investigated the impact of varying density(835 to 823 kg/m³), PAHs (5.7 percent to 1 percent),cetane number (53 and 55), and T95 (355°C to 317°C)for diesel on emissions. Emissions from heavy-dutydiesel engines were found to be rather insensitive tochanges in diesel quality. Euro II-compliant passengercars and light-duty diesel engines, by contrast, re-sponded more to decreasing density, PAHs, and T95.Reducing these three parameters from 835 kg/m³, 5.7vol%, and 355°C, respectively, to 823 kg/m³, 1 vol%,and 320°C lowered PM emissions by 20 percent forlight-duty vehicles but only 3.1 percent for heavy-duty vehicles. (These fuel parameter changes gave thelargest reductions in the study.) It is difficult to varydensity, polyaromatics, and T95 in diesel indepen-dently, and hence the reasons for these reductions arenot easy to identify readily.

Another consideration is that lowering densityand T95 would decrease the yield of middle distillates,an important observation when many developingcountries find it difficult to meet middle distillate de-mand while having a surplus of gasoline. Even in theEU, the combination of diesel parameters mentionedin the preceding paragraph made it difficult for therefining sector to meet the demand for middle distil-lates.

Current and Future StandardsIn order to support increasingly tighter emission stan-dards, fuel specifications have been made stringentover the years. Table A3.1 shows the evolution ofgasoline specifications in recent years. Lead wasbanned effective 1 January 2000 in the EU, but Greece,Italy, and Spain were granted until 2002 to phase leadout. The EU has not taken the same position regard-ing MTBE as the United States. There is no mandatefor the use of oxygenates in gasoline. In November

ANNEX 3. TRENDS IN VEHICULAR EMISSION STANDARDS AND FUEL SPECIFICATIONS IN THE EUROPEAN UNION 115

2000, the EU Working Group on the Classification andLabeling of Dangerous Substances examined the sta-tus of MTBE in a meeting of experts. The discussionresulted in the EU deciding that MTBE would not beclassified as a carcinogen, mutagen, or reproductivetoxin. The EU risk assessment, presented in January2001, concluded that MTBE is not a toxic threat tohealth, but that it can give drinking water a bad taste.In its draft directive for new fuel quality laws, pub-lished in May 2001, the EU set no limitations on theuse of MTBE in fuel after 2005. The EU has, however,recommended that MTBE be prevented from seepinginto groundwater through storage tanks at service sta-tions. As an example, the EU in September 2001 ap-proved a Danish tax initiative—tax breaks of €0.02per liter of gasoline sold at service stations fitted withleak-resistant underground tanks—to improve thequality of storage tanks.

Recent changes in EU diesel specifications areshown in table A3.2. Concerned about the effects ofsulfur in diesel on particulate control aftertreatmentdevices, Germany has taken steps to make diesel fuel

Automotive Gasoline Specifications in the EU

Parameter Unit 1993a 2000 2005 2009

Lead, maximum g/l 0.013 0.005 0.005 0.005Summer RVP kPa Various 60/70b 60/70b 60/70b

Benzene, maximum vol% 5.0 1.0 1.0 1.0Aromatics, maximum vol% — 42 35 35Olefins, maximum vol% — 18 18 18Oxygen, maximum wt% 2.7 2.7 2.7 2.7

Methanol, maximum vol% — 3.0 3.0 3.0Ethanol, maximum vol% — 5.0 5.0 5.0Isopropyl alcohol, maximum vol% — 10.0 10.0 10.0Tert-butyl alcohol, maximum vol% — 7.0 7.0 7.0Iso-butyl alcohol, maximum vol% — 10.0 10.0 10.0Ethers, C5+, maximum vol% — 15.0 15.0 15.0Other oxygenates, maximum vol% — 10.0 10.0 10.0

Sulfur, maximum wt ppm 500 150 50 10

— Limits not specified.

aSpecifications for unleaded gasoline.b60 except for member states with arctic or severe winter conditions for which a limit of 70 kPa applies.

Sources: Concawe 1997, EU 2003.

T A B L EA 3. 1

On-Road Diesel Specifications in the EU

1996Parameter Unit 1993 (Oct.) 2000 2005 2009

Sulfur, maximum wt ppm 2000 500 350 50 10Cetane number, 49 49 51 51 51 minimumCetane index, 46 46 — — — minimumT

95, maximum °C 370 370 360 360 360

PAHs, maximum % — — 11 11 11Density, maximum kg/liter 0.860 0.860 0.845 0.845 0.845

— Not applicable.

Source: Concawe 1997, EU 2003.

T A B L EA 3. 2

specifications even more stringent and set the maxi-mum limit to 10 ppm from January 2003. As a pre-liminary step, sulfur in diesel sold in Germany fromNovember 2001 had to be limited to 50 ppm. Ger-many has also urged other EU countries to adopt the10 ppm diesel sulfur standards in time for the sched-uled 1 October 2005 implementation date for Euro IV


truck and bus emission legislation. In early 2002, theEU Council of Ministers (Environment) agreed inprinciple that sulfur-free gasoline and diesel be intro-duced in every member state beginning in January2005, and that it be made mandatory by January 2009,two years earlier than the previously proposed date.

The exhaust emission limits for passenger carsand light commercial vehicles differentiated by thetype and weight of vehicle are given in table A3.3 andtable A3.4. EU emission standards for new light-dutyvehicles were originally specified in 1970. In June1991, the Council of Ministers of the European Com-munity adopted the Consolidated Emissions Direc-tive. Exhaust emission standards had to be certifiedon the basis of a combined United Nations EconomicCommission in Europe (ECE) 15 (urban) cycle andEUDC (extra-urban test cycle). In March 1994, theCouncil of Ministers adopted another directive withtightened limits from 1996 onwards in which separatestandards were given for gasoline- and diesel-fueledvehicles. The most recent limits for the so-called EuroIII and Euro IV were issued in 1998 in Directive 98/69/EC.

Taking effect from 1 January 2000, for new typesand from 1 January 2001, for all types, gasoline ve-

EU Exhaust Emission Standards forPassenger Cars (g/km)

Tier Year CO HC HC+NOx

NOx

PM

DieselEuro I 1992 2.72 — 0.97 — 0.14Euro II–IDI 1996 1.0 — 0.70 — 0.08Euro II–DI 1996–99 1.0 — 0.90 — 0.10Euro III 2000 0.64 — 0.56 0.50 0.05Euro IV 2005 0.50 — 0.30 0.25 0.025

GasolineEuro II 1996 2.2 — 0.50 — —Euro III 2000 2.3 0.2 — 0.15 —Euro IV 2005 1.0 0.1 — 0.08 —

— Not applicable; IDI indirect injection; DI direct injection.

Source: Concawe 1997.

T A B L EA 3. 3

EU Emission Standards for Light CommercialVehicles (g/km)

TierClass (Euro) Year CO HC HC+NO

xNO

xPM

Diesel1 I 1994 2.72 — 0.97 — 0.14

II–IDI 1998 1.0 — 0.70 — 0.08II–DIa 1998 1.0 — 0.90 — 0.1III 2000 0.64 — 0.56 0.5 0.05IV 2005 0.50 — 0.30 0.25 0.025

2 I 1994 5.17 — 1.4 — 0.19II–IDI 1998 1.25 — 1.0 — 0.12II–DIa 1998 1.25 — 1.3 — 0.14III 2002 0.80 — 0.72 0.65 0.07IV 2006 0.63 — 0.39 0.33 0.04

3 I 1994 6.9 — 1.7 — 0.25II–IDI 1998 1.5 — 1.2 — 0.17II–DIa 1998 1.5 — 1.6 — 0.2III 2002 0.95 — 0.86 0.78 0.1IV 2006 0.74 — 0.46 0.39 0.06

Gasoline1 I 1994 2.72 — 0.97 — —

II 1998 2.2 — 0.5 — —III 2000 2.3 0.2 — 0.15 —IV 2005 1.0 0.1 — 0.08 —

2 I 1994 5.17 — 1.4 — —II 1998 4.0 — 0.65 — —III 2002 4.17 0.25 — 0.18 —IV 2006 1.81 0.13 — 0.10 —

3 I 1994 6.9 — 1.7 — —II 1998 5.0 — 0.8 — —III 2002 5.22 0.29 — 0.21 —IV 2006 2.27 0.16 — 0.11 —

IDI Indirect injection; DI direct injection

aAfter September 30, 1999, DI engines are required to meet IDIengine limits.

Notes: For Euro I and II, the weight classes were Class 1 (<1250kg), Class 2 (1250–1700 kg), and Class 3 (>1700 kg). For Euro IIIand IV, Class 1 <1305 kg, Class 2 1305–1760 kg, and Class 3>1760 kg.

Source: Dieselnet undated.

T A B L EA 3. 4

hicles used for carrying passengers with no more thaneight seats in addition to the driver’s seat (the so-called M1 category)—except vehicles with a maxi-

ANNEX 3. TRENDS IN VEHICULAR EMISSION STANDARDS AND FUEL SPECIFICATIONS IN THE EUROPEAN UNION 117

mum mass exceeding 2,500 kg—and vehicles of class1 above are required to be fitted with an on-board di-agnostic (OBD) system. Taking effect from 1 January2001, for new types and from 1 January 2002, for alltypes, vehicles of classes 2 and 3 above and vehiclesof category M1 exceeding 2,500 kg must be fitted withan OBD system (European Parliament 2000).

For diesel vehicles, taking effect from 1 January2003, for new types and from 1 January 2004, for alltypes, category M1 vehicles—except vehicles with amaximum mass exceeding 2,500 kg or vehicles de-signed to carry more than six occupants including thedriver—must be fitted with an OBD system. All othertypes of M1 category vehicles and new types of class1 vehicles above must be fitted with an OBD systemtaking effect from 1 January 2005. New types of class2 and class 3 categories above are required to be fittedwith an OBD system taking effect from 1 January2006 (European Communities 1998).

Emission standards for heavy-duty and other en-gines are given in table A3.5 and table A3.6. In exam-ining vehicle emission standards in the EU, it is im-portant to bear in mind that EU test driving cycles’reference fuels are different from those in the UnitedStates, so that the numerical emission limits are notstrictly comparable. This holds especially in the case

Emission Standards for Diesel and GasEngines, European Transient Cycle Test(g/kWh)

Date andTier category CO NMHC CH

4a NO

xPMb

Euro III Oct. 1999, 3.0 0.4 0.65 2.0 0.02EEVs onlyJan. 2000 5.45 0.78 1.6 5.0 0.16

0.21c

Euro IV Jan. 2005 4.0 0.55 1.1 3.5 0.03Euro V Jan. 2008 4.0 0.55 1.1 2.0 0.03

NMCH Nonmethane hydrocarbons; CH4 methane; EEV environ-mentally enhanced vehicle.

aFor natural gas engines only.bNot applicable for gas fueled engines at the year 2000 and 2005stages.cFor engines of less than 0.75 cubic decimeter swept volume percylinder and a rated power speed of more than 3,000 per minute.


T A B L EA 3. 6

EU Emission Standards for Heavy-Duty Diesel Engines (g/kWh)

Tier Date and category Test cycle CO HC NOx

PM Smokea

Euro I 1992, <85 kW ECE R-49 4.5 1.1 8.0 0.612 —1992, >85 kW ECE R-49 4.5 1.1 8.0 0.36 —

Euro II Jan. 1996 ECE R-49 4.0 1.1 7.0 0.25 —Jan. 1998 ECE R-49 4.0 1.1 7.0 0.15 —

Euro III Oct. 1999, EEVs only ESC and ELR 1.5 0.25 2.0 0.02 0.15Jan. 2000 ESC and ELR 2.1 0.66 5.0 0.10 0.8

0.13b

Euro IV Jan. 2005 ESC and ELR 1.5 0.46 3.5 0.02 0.5Euro V Jan. 2008 ESC and ELR 1.5 0.46 2.0 0.02 0.5

kW Kilowatts; — not applicable; EEV environmentally enhanced vehicle; ESC European stationary cycle; ELR European load response.

aSmoke in m-1.bFor engines of less than 0.75 cubic decimeter swept volume per cylinder and a rated power speed of more than 3000 min-1.


T A B L EA 3. 5

of heavy-duty vehicles. The EU did not adopt a tran-sient driving cycle, which is more stringent than asteady-state mode for particulate emissions, until2000, whereas the U.S. EPA mandated a transient testin 1985, replacing a steady-state 13-mode test(Concawe 1997).

119

World-Wide Fuel Charter

The World-Wide Fuel Charter was developed by fourgroups—the Association des ConstructeursEuropéens d’Automobiles (ACEA), Alliance of Auto-mobile Manufacturers, Engine Manufacturers Asso-ciation (EMA), and Japanese Automobile Manufac-turers Association—and some of their affiliates. Theobjective of the global fuel harmonization effort, ac-cording to the auto makers, is to develop common,worldwide recommendations for “quality fuels,” tak-ing into consideration customer requirements and ve-hicular emissions technologies that will, in turn, ben-efit customers and all other affected parties. The automakers hope that implementation of the recommen-dations will (1) reduce vehicular emissions, (2) consis-tently satisfy customer performance expectations, and(3) minimize vehicle equipment complexities with op-timized fuels for each emissions control category.

The charter establishes four categories of un-leaded gasoline and diesel. Category 1 is intended formarkets that require minimal emissions controls. Cat-egory 2 is for markets with stringent requirements foremissions controls. Category 3 fuels are designed formarkets with advanced requirements for emissionscontrol as these technologies are designed today. Cat-egory 4, which was not included in the original char-ter but was added in the April 2000 edition, is formarkets with further advanced requirements foremission control (such as Euro IV and U.S. Tier 2) toenable sophisticated NOx and particulate matteraftertreatment technologies. The only difference be-tween category 3 and category 4 is the level of sulfur.

Some of the key features of gasoline fuel proper-ties are given in table A4.1. The charter states thatwhere oxygenates are used, ethers are preferred. Thelimit on oxygen corresponds to a maximum of 7.7vol% ethanol. Where up to 10 vol% ethanol is permit-ted by preexisting regulation, the blended fuel mustmeet all fuel requirements and fueling pump labeling

is recommended. Methanol is not permitted, andhigher alcohols with more than 2 carbon atoms arelimited to a maximum of 0.1 vol%.

Key diesel fuel properties are given in table A4.2.A maximum of 5 vol% fatty acid methyl esters (biodiesel)are permitted in categories 1–3 but not category 4.Ethanol and methanol must both be below the detec-tion limit for all four categories: that is, so-called e-diesel, a blend of ethanol and diesel, is not permitted.

The above specifications show the extent to whichthey are driven by the standards in industrial coun-tries. While most developing countries are in a posi-tion to meet category 1, there is a significant leap fromcategory 1 to category 2. For example, diesel sulfur islowered from 5,000 wt ppm to 300 wt ppm. In theevolution of diesel sulfur specifications, industrialcountries did not take such a large step. While somecountries may find it cost-effective to move immedi-ately to low or ultralow sulfur levels, others (such asthe case of Sri Lanka discussed in annex 1) couldmove from 5,000 wt ppm to 3,000 or 2,500 wt ppm atlittle cost, but find it prohibitively expensive to movefrom 5,000 to 350 wt ppm. Inserting several categoriesbetween category 1 and category 2 to address the needsand reality of developing countries might be useful, butwould also go against the objective of minimizing thenumber of categories. This illustrates the dilemma in har-monizing standards across the world against the back-ground of varying environmental goals, air quality prob-lems, climatic and geographical conditions, resourceendowments, and refinery configurations.

To the extent that harmonization can be beneficial,fuel specifications should be harmonized with emis-sion limits, vehicle technology, and test cycles. Not do-ing so and going after category 2 or category 3 fuel speci-fications, or worse, selecting just one fuel parameter inthe fuel specifications (such as diesel sulfur), could resultin the selection of suboptimal mitigation strategies.

A N N E X4


World-Wide Fuel Charter Gasoline Specifications

Gasoline parameter Unit Category 1 Category 2 Category 3 Category 4

91 RON, minimum RON 91.0 91.0 91.0 91.0MON 82.5 82.5 82.5 82.5



Lead, maximum g/liter 0.40a Below detectionb

Sulfur, maximum wt ppm 1,000 200 30 Sulfur-freec

Oxygen, maximum wt% 2.7 2.7 2.7 2.7Olefins, maximum vol% No limit 20 10 10Aromatics, maximum vol% 50 40 35 35Benzene, maximum vol% 5.0 2.5 1.0 1.0Density kg/m³ 715–780 715–770 715–770 715–770

RON Research octane number; MON motor octane number.

aWhere lead is legally permitted.bAt or below detection limit of test method used, no intentional addition.c5–10 wt ppm based on available data on advanced technology vehicles. As more data become available, a more specific maximum will be de-fined.

Category 2 for markets with stringent requirements for emission controls or other market demands.Category 3 for markets with advanced requirements for emission controls or other market demandsCategory 4 for markets with further advanced requirements for emission control, to enable sophisticated NOx technologies.

Source: ACEA and others 2002.

T A B L EA 4. 1

World-Wide Fuel Charter Diesel Specifications

Diesel parameter Unit Category 1 Category 2 Category 3 Category 4

Cetane number, minimuma 48 53 55 55Cetane index, minimuma 45 50 52 52Density at 15°C kg/m³ 820–860 820–850 820–840 820–840Kinematic viscosity at 40°C cSt 2.0–4.5 2.0–4.0 2.0–4.0 2.0–4.0Sulfur, maximum wt ppm 5,000 300 30 Sulfur-freeb

Aromatics, maximum wt% Not specified 25 15 15PAHs, maximum wt% Not specified 5 2 2T

90, maximum °C Not specified 340 320 320

T95

, maximum °C 370 355 340 340

cSt Centistokes.

aCompliance with either cetane index or cetane number is allowed.b5-10 wt ppm based on available data on advanced technology vehicles. As more data become available, a more specific maximum will be de-fined.

Source: ACEA and others 2002.

T A B L EA 4. 2

121

Two- and Three-Wheelers

Emissions from the large and rapidly growing num-ber of two- and three-wheel vehicles are a majorsource of air pollution in a number of countries, espe-cially in Asia. Because they are less expensive thanother vehicles, two- and three-wheelers play an im-portant role in the transport market. In South Asia,they accounted for at least half of all vehicles as of2000. Two-wheel vehicles, which include mopeds,scooters, and motorcycles, are used mostly for per-sonal transportation, although in some cities (for ex-ample, Bangkok) they are used as taxis. Three-wheelvehicles, which include small taxis such asautorickshaws and larger vehicles that hold as manyas a dozen passengers, are used commercially. In theearly 1990s, the majority of three-wheelers and two-wheelers in Asia had two-stroke engines.

Conventional two-stroke engines have several ad-vantages over four-stroke engines. These includelower cost, excellent power, mechanical simplicity(fewer moving parts and resulting ease of mainte-nance), lighter and smaller engines, greater operatingsmoothness, lower crankcase emissions throughoutthe life of the vehicle, and lower NOx emissions (be-cause two-stroke engines run so rich that NOx forma-tion is suppressed). They also have disadvantagescompared with four-stroke gasoline engine vehicles,including higher particulate and hydrocarbon emis-sions, lower fuel economy, and louder noise.

Until recently new two-stroke engines withoutcatalysts emitted as much as an order of magnitudemore particulate matter than four-stroke engines ofsimilar size. When vehicle age, maintenance, lubri-cant, and fuel quality are taken into account, two-stroke engines in developing countries probably emitparticulate matter at an even higher factor. As muchas 15 to 35 percent of the fuel-air mixture from two-stroke engine vehicles escapes through the exhaustport. These “scavenging losses” contain a high level

of unburned gasoline and lubricant. Some of the in-completely burned lubricant and heavier portions ofgasoline are emitted as small oil droplets that increasevisible smoke and particulate emissions.

Recently vehicle manufacturers in industrial coun-tries have been turning to direct-injection small two-stroke engines to meet extremely tight exhaust emis-sion standards. There are concerns that these enginesmay increase the number (as opposed to mass) of par-ticles emitted. Since environmental legislation is in-creasingly turning to the importance of controllingparticle size and number, this question needs to be in-vestigated further.

While particulate emissions from traditional two-stroke engines are high, it is worth bearing in mindthat particles emitted by two-stroke gasoline enginesare fundamentally different from those emitted bydiesel vehicles. The health impact of oil droplet-basedparticles from two-stroke engine vehicles is not wellunderstood. Most health impact studies have beencarried out in countries that do not have large two-stroke engine vehicle populations and where the prin-cipal sources of fine particulate emissions are dieselvehicles and stationary sources. In all studies, sicknessand death are regressed on the overall ambient par-ticulate concentrations, not against vehicular particu-late emissions. Most of the particulate matter fromtwo-stroke engines is soluble organic matter, whereasparticulate matter from diesel vehicles and stationarysources contains a significant amount of graphitic car-bon. Their behavior in the atmosphere in terms ofnucleation, agglomeration, dispersion, and condensa-tion could be quite different. This area of researchmerits further investigation.

The age and poor maintenance of many two- andthree-wheelers in developing countries increase emis-sions well above any applicable standards. In addi-tion, many drivers use lubricants of poor quality,

A N N E X5


leading to two distinct but related problems (Kojimaand others 2000):

� Drivers in some countries use widely availablestraight mineral oil—intended for use only inslow-moving stationary engines but not in ve-hicles—or new or recycled four-stroke engine oilrather than the specially formulated 2T oil rec-ommended by vehicle manufacturers. These oilsbuild up deposits in the engine, increasing emis-sions.

� Drivers may also use excessive quantities of lu-bricant. Some drivers simply lack knowledgeabout the correct amount of lubricant to addand the adverse effects of using too much. Oth-ers believe that adding extra lubricant improvesvehicle performance, increases fuel economy, orprovides greater protection against piston sei-zure. Because straight mineral oil may not mixwith gasoline as well as 2T oil, a greater quan-tity of lubricant is needed for lubrication. In re-ality this practice provides little or no benefit todrivers, but significantly increases the level ofemissions and reduces the quality of air for soci-ety at large.

Drivers are also encouraged to buy much more lu-bricant than needed by filling station operators, whoearn higher margins on oil than on gasoline. Further-more, adulteration of gasoline with kerosene is wide-spread in a number of developing countries becauseof the large difference in the retail price of the two fu-els. This practice increases emissions because kero-sene has a higher boiling point than gasoline and istherefore more difficult to burn. As a result more de-posits build up in the engine and damage the engineover time, and more unburned hydrocarbons andparticles are emitted in the exhaust gas.

Relationships between Mass Emissionsand Vehicle and Fuel/LubricantTechnologyOf special concern to policymakers are emission char-acteristics of in-use vehicles. Type approval and con-formity of production tests are carried out on new ve-

hicles using certified reference fuels and lubricants.With increasing vehicle usage, lack of maintenance,and perhaps use of improper lubricants or adulter-ated gasoline, vehicle emission levels can be several-fold higher.

One recent study (Kojima and others 2002,ESMAP 2002a) measured mass emissions from threetwo-stroke engine gasoline three-wheelers commer-cially operated in Dhaka, Bangladesh. Although it isnot possible to obtain statistically meaningful resultswithout conducting tests on a large fleet, the studyprovides some stylized observations. The three ve-hicles were manufactured by Bajaj Auto, India, in1993 (vehicle 3), 1995 (vehicle 1), and 1996 (vehicle 2).Vehicle 2 was manufactured to meet the 1996 Indianemission standards. The parameters investigated fortheir impact on pollutant emissions were the vehicleage (4, 5, and 7 years); state of vehicle maintenance;quality of gasoline (1998 “reference” gasoline pur-chased in India with 87 research octane number[RON], and a blend of gasoline from five filling sta-tions in Dhaka with 80 RON; gasoline sold in Dhakais often adulterated with kerosene); quality of lubri-cant (straight mineral oil used in Dhaka by three-wheeler drivers, 2T oil meeting JASO FB1 specifica-tions purchased in India, and JASO FC grade “lowsmoke” oil purchased in India); quantity of lubricant(3 percent as recommended by the vehicle manufac-turer, and 8 percent representing the common practicein Dhaka of using excessive quantities of lubricatingoils); and the installation of an oxidation catalyst(manufactured by Allied Signal).

Vehicles 1 and 3 were tested before and after ser-vicing, which consisted of cleaning or adjusting thespark plug, carburetor, air filter, and clutch/gear/ac-celerator play, and correcting for leakage in the ex-haust system. The test sequence was randomized toreduce the impact of systematic errors. In order to fur-

1In 1990, the Japanese Standards Organization (JASO) createda two-stroke lubricant standard with three levels of quality (FA,FB, and FC). Lubricity and detergency quality increase from FA toFC, and exhaust blocking and smoke emission improve. Maxi-mum permissible levels of smoke density are 50 percent for FA oil,44 percent for FB oil, and 24 percent for FC oil. Japanese manufac-turers of two-stroke vehicles identify FC (low-smoke lubricants) astheir minimum requirement.

ANNEX 5. TWO- AND THREE-WHEELERS 123

ther provide a check on systematic errors, vehicle 2was designed to be tested without undergoing a ser-vice. Unfortunately, vehicle 2 had a major breakdownafter six tests and had to be repaired. The repair car-ried out on vehicle 2 was intended to represent mini-mal “reconditioning” of the vehicle and not the ser-vice performed on vehicles 1 and 3. Only the seizedpiston and rings were replaced with utmost carewhile ensuring that the engine deposits and so onwere not disturbed. However, the results obtainedshowed that even this reconditioning lowered emis-sions. The results were analyzed using multiple linearregression analysis. The explanatory variables areshown in table A5.1. The “base case” was selected tobe vehicle 2 after reconditioning without the catalystusing the Indian reference gasoline and 3 percent 2Toil. This combination gives one of the lowest particu-late emission levels.

The results for particulate emissions are given intable A5.2. The mass particulate emissions from threein-use three-wheelers operating in Dhaka were high,ranging from 0.16 to 2.7 g/km, and averaging 0.7 g/km. Mass particulate emissions followed the expectedtrend with the exception of vehicle service—using

Independent Variables in Regression Analysis

Parameter Description

V1 1 if the vehicle is 1 before service, 0 otherwiseV2 1 if the vehicle is 2 before reconditioning, 0

otherwiseV3 1 if the vehicle is 3 before service, 0 otherwiseV1S 1 if the vehicle is 1 after service, 0 otherwiseV3S 1 if the vehicle is 3 after service, 0 otherwiseL1 1 if the lubricant is straight mineral oil, 0

otherwiseL3 1 if the lubricant meets JASO FC specifications,

0 otherwiseQ2 1 if the lubricant quantity is 8%, 0 if 3%G1 1 if gasoline is from Dhaka, 0 if it is 87 RON

“reference” gasoline from IndiaCat 1 if vehicle 2 is equipped with an oxidation

catalyst, 0 otherwiseSN Numbers ranging from 1 to 43 corresponding to

the test number in the experimental program

T A B L EA 5. 1

straight mineral oil, increasing the concentration of lu-bricant added, and using gasoline purchased inDhaka (which, as noted above, is often adulteratedwith kerosene) increased emissions. The base casegave particulate emissions of 0.19 g/km. The emis-sions from vehicle 3 before repair were nearly fivetimes higher than those for vehicle 2 after recondition-ing. Repairing vehicle 3 reduced particulate emissionsby less than 30 percent, and the mass emissions re-mained substantially higher than those of unservicedvehicles 1 and 2. Although vehicle 2 went throughonly reconditioning rather than a full service, thisseems to have had a considerable impact on particu-late emissions, resulting in a decrease of 40 percent. Inthe case of vehicle 1, repairing actually increasedemissions by one-third. Increasing the concentrationof lubricant from 3 to 8 percent increased emissionsby 61 percent. Using straight mineral oil instead of 2Toil increased emissions by about the same amount.Using gasoline from Dhaka rather than the gasolinepurchased in India increased emissions by 14 percent.

An interesting finding is that while lubricantsmeeting JASO FC specifications are known to pro-duce much less visible smoke than regular 2T oilbased only on mineral oil, the so-called low-smoke oilshowed no improvement in terms of mass particulateemissions. It is clear from the above findings that themechanical condition of the vehicle has the most sig-nificant impact on particulate emissions.

The results of HC and CO emissions are given intable A5.3 and table A5.4, respectively. There was es-sentially no impact of the quality of lubricant. Increas-ing the concentration of lubricant lowered HC andCO emissions, as did the use of gasoline from Dhaka.

The oxidation catalyst halved HC emissions, low-ered particulate emissions by about one-third, and re-duced CO emissions by about 15 percent. There wassome correlation between smoke measurements andmass particulate emissions, but the relationship wasweak. The correlation between smoke and mass par-ticulate emissions was poor below 1 g/km. For idlesmoke, there was a reasonable relationship above par-ticulate emissions of 1 g/km. Under free acceleration,the smoke meter reached saturation above 1 g/km.


Log Particulate Emission Model Specification

Variable Coefficient Standard error t-Statistic 10(coefficient)

Constant -0.72 0.028 -25.9 0.19V2 (vehicle 2) 0.14 0.037 3.9 1.39V3 (vehicle 3) 0.69 0.034 20.4 4.92V1S (V1 after service) 0.13 0.038 3.3 1.34V3S (V3 after service) 0.55 0.035 15.7 3.53G1 (Dhaka gasoline) 0.06 0.024 2.4 1.14L1 (straight mineral oil) 0.21 0.026 8.3 1.63Q2 (8% lubricant) 0.21 0.024 8.8 1.61

Mean dependent variable -0.26 Durbin-Watson statistic 2.33Dependent variable standard deviation 0.33 Jarque-Bera statistic 1.79R-squared 0.96 F-statistic 109

T A B L EA 5. 2

Log HC Emission Model Specification

Variable Coefficient Standard error t-Statistic 10(coefficient)

Constant 0.96 0.008 114.3 9.03V3 (vehicle 3) 0.48 0.012 41.1 3.02V3S (V3 after service) 0.31 0.012 26.2 2.02L3 (low smoke oil) 0.02 0.009 2.1 1.04Q2 (8% lubricant) -0.02 0.008 -2.9 0.95G1 (Dhaka gasoline) -0.08 0.008 -8.9 0.84

Mean dependent variable 1.1 Durbin-Watson statistic 2.15Dependent variable standard deviation 0.2 Jarque-Bera statistic 6.69R-squared 0.99 F-statistic 451

T A B L EA 5. 3

CO Emission Model Specification

Variable Coefficient Standard error t-Statistic

Constant 15.5 0.53 29.0V1 (vehicle 1) -6.50 0.81 -8.1V2 (vehicle 2) -6.35 0.71 -8.9V3 (vehicle 3) 10.13 0.67 15.1V1S (V1 after service) -10.70 0.75 -14.2V3S (V3 after service) 1.78 0.68 2.6Q2 (8% lubricant) -1.54 0.44 -3.5G1 (Dhaka gasoline) -1.03 0.45 -2.3

Mean dependent variable 13.4 Durbin-Watson statistic 2.21Dependent variable standard deviation 6.7 Jarque-Bera statistic 1.31R-squared 0.97 F-statistic 128

T A B L EA 5. 4


Since these trends may be a function of the instrumentused to measure smoke (Celesco Model 300 PortableOpacity Meter, produced by Telonic Berkeley Inc.,California, United States, was used in this study),more work should be carried out to check this correla-tion. If the correlation is indeed poor, this brings intoquestion the merit of setting low smoke emission lim-its for in-use two-stroke engine vehicles other than tocheck lubricant use.

Emission Standards for Two- and Three-Wheel VehiclesDespite concerns about high particulate emissionsfrom two-stroke engines, no country has mass par-ticulate emission standards for two- and three-wheel-ers. The United States, Europe, and Japan did nothave tight emission standards for two- and three-wheel vehicles in the 1990s. In fact, Japan did nothave any emission standards for two-wheelers formost of the 1990s and adopted exhaust emission lim-its for the first time in 1998/1999. Areas in Asia withlarge two- and three-wheel populations, such as Indiaand Taiwan, China, have historically adopted muchmore stringent emission standards.

U.S. Emission Limits for Motorcycles over 50 cc Capacity

Regulation Model year Engine capacity D (cc) CO (g/km) HC (g/km)

Federal and California 1978 and subsequent 50 to less than 170 17.0 5.0170 to less than 750 17.0 5.0 + 0.0155×(D-170)750 or greater 17.0 14.0

California only 1980–81 50 or greater 17.0 5.01982 and subsequent 50–279 12.0 1.01982–1985 280 or greater 12.0 2.51985–1987 280 or greater 12.0 1.0a

1988–2003 280–699 12.0 1.0a

1988–2003 700 or greater 12.0 1.4a

2004–2007 280 or greater 12.0 1.4a, b

2008 and subsequent 280 or greater 12.0 0.8a, b

aCorporate average.bHC + NOx.

Sources: Concawe 1997, CARB 2000b.

T A B L EA 5. 5

The U.S. EPA adopted emission limits for motor-cycles in 1978 but excluded motorcycles with enginessmaller than 50 cubic centimeters (cc) (table A5.5).California has steadily tightened emission standardsfor motorcycles. The EPA in July 2002 proposed re-vised emission standards, and included standards forpreviously unregulated two-wheelers with enginessmaller than 50 cc such as scooters and mopeds (tableA5.6). The EPA expects the class III standards to bemet by an increased use of technologies already dem-onstrated as being effective in four-stroke motorcycleengines, such as secondary air injection, electronic

Future U.S. Motorcycle Exhaust EmissionStandards

Imple-Engine mentation HC HC+NOx CO

Class size (cc) date (g/km) (g/km) (g/km)

I <180 2006 1.0 — 12.0II 180–279 2006 1.0 — 12.0III >280 2006 — 1.4 12.0III >280 2010 — 0.8 12.0

— Not applicable.

Source: U.S. EPA 2003.

T A B L EA 5. 6


ECE Regulation 40/40.01 for Type Approval Exhaust Emission Limits for Four-Stroke Engine Motorcycles

Reference weight CO (g/km) HC (g/km) Ra (kg) ECE 40 ECE 40.01 ECE 40 ECE 40.01

Four-stroke<100 25 17.5 7 4.2100–300 25+25/200×(R-100) 17.5+17.5/200×(R-100) 7+3×(R-100)/200 4.2+1.8×(R-100)/200>300 50 35 10 6

Two-stroke<100 16 12.8 10 8100–300 16+24×(R-100)/200 12.8+19.2×(R-100)/200 10+5×(R-100)/200 8+4×(R-100)/200>300 40 32 15 12

aReference weight R = motorcycle weight + 75 kg.

Source: UNECE 2002.

T A B L EA 5. 7

fuel injection systems, and catalytic converters. Thestandards are not expected to result in the universaluse of catalytic converters. The EPA projects averagecosts of US$26 per motorcycle to meet the 2006 stan-dards and US$35 to meet the 2010 standards.

In Europe, ECE Regulation 40 was adopted in1979 and applies to two- and three-wheelers with anunladen weight of less than 400 kg and having amaximum design speed exceeding 50 km/h or an en-gine capacity greater than 50 cc. ECE Regulation 40was amended in 1988 to become ECE 40.01. TableA5.7 shows the emission limits that ECE 40 and 40.01required to be met in a driving cycle. There was also aCO idle emission limit of 4.5 vol%. These limits weremade more stringent in Directive 97/24/EC, whichbecame mandatory for new EU type approvals begin-ning in June 1999 (table A5.8).

The above limits will be tightened further in 2003(table A5.9). Unlike the existing standards, the newproposal does not distinguish between two- and four-stroke engines. Previous emission limits for tricyclesand quadricycles were set at roughly 1.5 times thoseof two-wheeled motorcycles, but the new proposedstandards for those vehicles have been set at about1.25 times the proposed motorcycle standards. TheEU allowed the use of fiscal incentives to encouragethe early introduction of the 2003 standards. It also in-troduced permissive values for member states wish-ing to encourage emission standards beyond those

EU Motorcycle Emission Limits, 1999–2003

Type CO (g/km) HC (g/km) NOx (g/km) Cycle

Two-stroke 8.0 4.0 0.1 ECE-40Four-stroke 13.0 3.0 0.3 ECE-40

Source: European Commission 2000b.

T A B L EA 5. 8

Common Position on Motorcycle EmissionStandards Adopted in July 2001 by theEuropean Council (g/km)

Legal limit Voluntary limitsfrom 2003 for tax incentives

Engine size CO HC NOx

CO HC NOx

<150 cc 5.5 1.2 0.3 2 0.8 0.2>150 cc 5.5 1.2 0.3 2 0.3 0.1

Source: European Commission 2000b.

T A B L EA 5. 9

previewed for 2003 through the use of fiscal incen-tives. For three-wheelers, the limits are 7.0, 1.5, and0.4 g/km for CO, HC, and NOx, respectively, also be-ginning in 2003.

India and Taiwan, China, are among the worldleaders in setting stringent emission standards formotorcycles. Their respective standards are shown intable A5.10 and table A5.11. Although the CO limitsare numerically higher in 2003 in Taiwan, China, than


Type Approval Emission Standards forGasoline- and Diesel-Powered Two- and Three-Wheelers in India (g/km)

Two- andTwo- Three- three-

wheelers wheelers wheelersYear CO HC+NO

xCO HC+NO

xPM

1991 12–15a 8–9a, b 30 12b —1996 4.5 3.6 6.75 5.4 —1998 4.5 3.6 6.75 5.4 —2000 gasoline 2.0 2.0 4.0 2.0 —2000 diesel 2.72 0.97 2.72 0.97 0.142005 gasoline 1.5c 1.5c 2.25c 2.0c —2005 diesel 1.0d 0.85e 1.0d 0.85e 0.10c

— Not applicable.

aEmission standard depends on the reference mass of the vehicle.bLimit applied to HC only and not to HC+NOx.cA deterioration factor of 1.2 applies to the first 30,000 km.dA deterioration factor of 1.1 applies to the first 30,000 km.eA deterioration factor of 1.0 applies to the first 30,000 km.

Notes: Tests of 1991 and 1996 vehicles were based on the warmIndian driving cycle. Tests of 1998, 2000, and 2005 are based onthe cold Indian driving cycle. Test procedures for durability require-ments for the year 2005 standards have not yet been notified as ofApril 2004. Until their notification, the actual emission levels of newvehicles are required to be lower than the stipulated limits by thecorresponding deterioration factor.

Source: Iyer 2004.

T A B L EA 5. 1 0

in 1998, the requirement to test the vehicle using thecold-start ECE 15 driving cycle means that the stan-dards are more stringent. Setting tighter standards in2003 for two-stroke than for four-stroke motorcycleengines effectively rules out the use of conventional

Emission Standards for New Production Motorcycle Models in Taiwan, China

Stage Category Effective date Cycle HC+NOx (g/km) CO (g/km) Durability (km)

1 All 1 Jan. 1988 ECE 40 5.5 8.8 None2 All 1 July 1991 ECE 40 3.0 4.5 6,0003 All 1 Jan. 1998 ECE 40 2.0 3.5 15,0004 Two-stroke 1 Jan. 2004 ECE 15 cold start 1.0 7.0 15,0004 Four-stroke 1 Jan. 2004 ECE 15 cold start 2.0 7.0 15,000

Note: The limits shown for 2004 correspond to motorcycles with an engine capacity smaller than 700 cc.

Source: Shu 2001.

T A B L EA 5. 11

two-stroke engines. The authorities introduced dura-bility requirements in 1991, set at 6,000 km, whichwere increased in 1998 to 15,000 km. The year 2000Indian limit on CO of 2 g/km for two-wheelers istighter than the EU limit of 5.5 g/km, which cameinto force in 2003. Similarly the year 2000 Indian limiton CO of 4.0 g/km for three-wheelers is tighter thanthe EU limit of 7.0 g/km. For the 2005 emission stan-dards, the government of India, for the first time, setdeterioration factors applicable for the first 30,000 kmfor two- and three-wheelers. This is twice the distancestipulated in the durability requirements in Taiwan,China. Furthermore, for the first time, the limits fortype approval and conformity of production weremade identical, making the limits much more strin-gent in practice than they appear in table A5.10.

Controlling Emissions from Two- andThree-WheelersEmissions from the existing fleet of two-stroke gasolineengines can be reduced by (a) ensuring that driversuse the correct type and quantity of lubricant, (b) im-proving vehicle maintenance, and (c) stopping theadulteration of gasoline with higher boiling point ma-terials such as kerosene. For new vehicles, switching tofour-stroke engine vehicles at the time of vehicle re-placement would be very cost-effective. Other optionsinclude adopting direct injection two-stroke enginetechnology or using alternative fuels such as LPG,which is used extensively in three-wheelers inBangkok, and CNG, which is used in three-wheelersin India and Pakistan.


Promoting the correct use of lubricant in existingtwo-stroke engine vehicles requires mass education ofdrivers, filling station attendants, vehicle owners,regulators, and even the public, which has a role inbringing political pressure to bear on the problem. Incities where drivers are using excess lubricant, cuttingback to the correct amount recommended by manu-facturers (3 percent or lower, depending on the ve-

hicle type) could save drivers money even if theyswitch to superior quality (and hence more expen-sive) lubricant. Even though the oil itself is more ex-pensive, analysis often shows that 2T oil used in theproper amount costs less than drivers currently payfor larger amounts of lower-grade oils. In additionthere are difficult-to-quantify benefits of longer en-gine life and lower emissions.

129

Alternative Fuels

Vehicles fueled by gasoline and diesel can be veryclean if equipped with advanced exhaust- treatmentdevices that are maintained well to ensure durabilityand if fuels of required quality are used. However, itis expensive to deploy advanced-gasoline (for ex-ample, the equivalent of U.S. Tier 2) and advanced-diesel vehicles, and the requisite quality fuels, espe-cially with regard to sulfur, are not available in manydeveloping countries. Another way of reducing emis-sions is to switch to inherently cleaner fuels. Gaseousfuels in particular burn cleaner than liquid fuels do,and the combustion of hydrogen, as well as the use ofelectricity, does not emit any harmful pollutants at thetailpipe.

This annex presents an overview of some of themore commonly used alternative fuel vehicles, orthose that are considered to offer the greatest futurepotential. The fuels selected are natural gas, liquefiedpetroleum gas (LPG), electricity, biofuels, and hydro-gen (used in fuel cell technology). There is alreadyquite a bit of worldwide experience with natural gas,LPG, and ethanol. The experiences with alternativefuels seem to suggest a longer transition away fromconventional fuels than initially anticipated.

Natural GasNatural gas (NG) offers a significant potential for re-ducing harmful emissions from vehicles, especiallythose of fine particles, compared to conventional (asopposed to advanced or “clean” diesel with ultralowsulfur and employing advanced control technology)diesel. As of late 2003, there were more than 3.3 mil-lion natural gas vehicles (NGVs) and 6,600 refuelingstations worldwide. More than half of the total natu-ral gas vehicle population was in Argentina and Bra-zil (IANGV undated). By far the majority of naturalgas vehicles are gasoline vehicles converted to com-

pressed natural gas (CNG). The alternative of storingnatural gas in the liquid state, liquefied natural gas(LNG), is much less common, although the liquefac-tion process removes impurities and gives higher pu-rity natural gas (higher methane content) than CNG.

Methane, which constitutes the bulk of CNG, hasan antiknock index (average of research and motoroctane numbers) of over 120. Dedicated CNG vehiclescan take advantage of this high octane number of thefuel and operate at a high compression ratio. In prac-tice, the composition of pipeline natural gas varies de-pending on the source and processing of the gas, aswell as the time of year. As a result, not only does thefuel octane number vary, but also the heating valuecan vary by as much as 25 percent, affecting vehicleperformance. Moreover, when used as a fuel in ve-hicles, the heavier hydrocarbons in natural gas cancondense, and the condensation and revaporizationlead to fuel-enrichment variations that affect bothemissions and engine performance. The presence ofhigher hydrocarbons and compressor oils can causeknock. Natural gas should be stripped of heavier hy-drocarbons before distribution to avoid this problem.

There are two primary reasons for switching tonatural gas:

1. Diversification of energy sources has been thehistorical reason for selecting natural gas as amotor fuel. At the end of 2002, the ratio ofproven reserves to production of natural gaswas estimated to be 61 years, 50 percent higherthan that of oil at 41 years (bp 2003).

2. Much lower emissions, especially comparedwith conventional diesel vehicles. This is the pri-mary reason for switching from diesel to naturalgas today.

In terms of fuel supply, there are three types ofNGVs:

A N N E X6


� Bi-fuel vehicles, which can run on either natu-ral gas or gasoline.

� Dual-fuel vehicles, which run on diesel only ordiesel and natural gas with the combustion ofdiesel used to ignite the natural gas. The stop-and-start nature of urban bus cycles limits thesubstitution of diesel by natural gas, and makesdual-fuel unsuitable if the objective is to reduceemissions (see box 4).

� Dedicated vehicles, which run entirely on natu-ral gas. There is a significant difference in tech-nology between dedicated NGVs that are sto-ichiometric—typically gasoline-derived engineswith a three-way catalyst—and “lean burn”built up from diesel blocks and designed tokeep NOx formation low.

All three types can be manufactured from the startto use natural gas by original equipment manufactur-ers (OEM) or converted from vehicles that were origi-nally manufactured to run on gasoline or diesel only.

Either way, there is an incremental cost in vehiclepurchase or conversion relative to vehicles using con-ventional liquid fuels, and this additional cost shouldideally be recovered from savings in operating costs,typically lower fuel costs. For minimizing emissions,OEM vehicles are considered superior to convertedones, but they are more expensive. Conversion of ve-hicles in poor condition, as well as poor conversions,are two of the most serious potential problems in de-

From Dual-Fuel Buses to DedicatedCNG: Lesson from Seattle, UnitedStates

Seattle put 129 CNG dual-fuel vehicles into service in 1992. Drivers,

however, were found not to like using CNG. In 1998, the city’s usage

of natural gas in the dual-fuel buses was discovered to be a mere 8

percent. An education campaign was mounted, and the natural gas

usage increased to 41 percent in response. The city is now choosing

dedicated CNG buses instead of dual-fuel vehicles, and these

single-fuel CNG buses are being deployed closer to refueling stations.

Source: Automotive Environment Analyst 2000.

B O X 4

veloping country cities, and could even compromisethe purpose of switching to NG.

The advantages of NGVs include

� Very low particulate emissions� Very low emissions of airborne toxins� Negligible sulfur-containing emissions� Quieter operation, having less vibrations and

less odor than the equivalent diesel engines.

Their disadvantages are as follows:

� Much more expensive fuel distribution and stor-age. Natural gas has to be transported through apipeline network and compressed to 200 timesthe atmospheric pressure or even more (forCNG) or liquefied to -162°C (for LNG). LNG’scomplex on-board storage system makes it suit-able only for heavy-duty vehicles.

� Higher vehicle cost, primarily due to the highercost of fuel cylinders.

� Shorter driving range. CNG and LNG containless energy per unit volume than gasoline ordiesel.

� Heavier fuel tank. This reduces fuel economyand leads to greater braking distance.

� Potential performance and operational prob-lems compared to liquid fuels (Watt 2001, Eudy2002). There are consistent reports that the per-formance of the generation of natural gas busesfrom the early 1990s was less than satisfactory.More recent models have shown much im-provement, but NG buses are not without prob-lems. The New York City Metropolitan Trans-portation Authority has been operating a fleet ofCNG buses, numbering 90 in 2000. Their experi-ence with CNG buses between 1995 and 2000showed, however, that CNG buses were only 50to 75 percent as reliable as diesel buses, 40 per-cent less energy-efficient in urban service, andsignificantly more expensive to operate (MTANew York City Transit Department of Buses2000).

With respect to emissions, it is worth noting thatadvanced technology gasoline vehicles with three-way catalysts are so clean that the fuel itself (that is,

ANNEX 6. ALTERNATIVE FUELS 131

whether liquid or gas) plays a minor role, especiallyfor the regulated emissions. Under these circum-stances, converting an advanced gasoline vehicle togaseous fuel could have little or no emissions benefits,with the exception of lower evaporative emissions ofnon-methane hydrocarbons.

NGVs have a marked advantage over conven-tional diesels. Example data taken from the UnitedStates comparing CNG with diesel shown in tableA6.1 amply illustrate this point.

Emissions Benefits of Replacing ConventionalDiesel with CNG

Fuel CO NOx PM

Diesel 2.4 g/km 21 g/km 0.38 g/kmCNG 0.4 g/km 8.9 g/km 0.012 g/km% reduction 84 58 97

Note: Medium-duty diesel buses, central business district testcycle.

Source: Frailey and others 2000.

T A B L EA 6. 1

A well-configured CNG bus produces lower NOx

emissions than a pre-2003 U.S. diesel bus, but post-2003 diesel buses are approaching the same low NOx

level through the use of exhaust gas recirculation(EGR). The emergence of so-called clean diesel, pilot-tested in North America and Europe and slated formandatory deployment by the latter half of this de-cade, poses a serious challenge to NGVs. As men-tioned in the first three annexes, most clean dieseltechnologies rely on dramatic reductions in the levelof sulfur in diesel to enable the use of such exhausttreatment devices as continuously regenerating par-ticulate filters and lean deNOx catalysts. Availabledata indicate that particulate emission levels betweenclean diesel and CNG may be essentially the same(table A6.2), although particulate emissions fromstate-of-the-art CNG vehicles (with a stoichiometricair-to-fuel ratio and a three-way catalyst) may still belower than those from state-of-the-art clean diesel ve-hicles. Note that table A6.2 shows total hydrocarbons,which include methane, and most of the hydrocarbonemissions from CNG buses comprise methane, which

is a powerful greenhouse gas but nonreactive and hasvirtually no impact on urban air pollution.

While many technical breakthroughs have beenannounced for the deployment of clean diesel, includ-ing refining processes to produce ultralow sulfur die-sel at a fraction of the cost employing conventionaltechnologies, clean diesel is likely to be some yearsaway for widespread application in a number of de-veloping countries, where natural gas remains theonly commercially tested clean fuel alternative forheavy-duty engine applications. Strong technical sup-port and awareness-raising are crucial for promotingthe switch from diesel to natural gas vehicles (see box5).

Two contrasting country cases: Argentina andNew ZealandArgentina and New Zealand were once two worldleaders in the NGV market. Today, Argentina’s mar-ket remains the largest in the world, while the NewZealand’s NGV market has declined precipitouslysince the late 1980s. The difference in experience ofthese two countries is instructive.

Argentina launched its CNG vehicle program in1984. By then, there was an extensive network ofnatural gas pipelines reaching most cities. The gov-ernment offered no direct subsidies; the incentive forfuel switching stemmed entirely from the high tax ongasoline. The fuel prices in December 1999 were

Comparison of CNG and “Clean-Diesel” Busesin New York (g/km)

Central business New Yorkdistrict cycle bus cycle

Pollutant CNG Diesel CNG Diesel

Particulate matter 0.011 0.015 0.044 0.023NOx 15 16 32 45Total hydrocarbons 10 0.01 42 0.038

Note: Heavy-duty diesel buses (1999 model year) using dieselcontaining 30 ppm sulfur and Johnson Matthey’s continuously re-generating particulate trap system (but not lean deNOx catalysts);CNG buses (1996, 1998, and 1999 model year) equipped with oxi-dation catalysts.

Source: MTA New York City Transit Department of Buses 2001.

T A B L EA 6. 2


Natural Gas Buses: Experience of BusFleet Operators

Phoenix Transit, USA. The bottom line is training, training, and more

training. Phoenix Transit initially met resistance from the operators,

mechanics, fuelers, and subsequently the union. The company

trained everyone, from top management to bus washers. The main

challenge is to have operators and maintenance staff who can ob-

serve and report changes in the buses while in operation and during

preventive maintenance. The next challenge is to have bus manufac-

turers and component manufacturers working in partnership with the

service and maintenance contractors.

Sydney buses, Australia. Perception problems about poor

drivability were put to rest with comparison trials with diesel buses.

The conclusion was that the lack of noise from the natural gas buses

gave the drivers the impression of lack of acceleration!

Source: Watt 2001.

B O X 5

US$1.04 per liter of premium gasoline, $0.50 per literof diesel, and $0.33 per cubic meter (m³) of CNG. Atthese prices, the payback period for those vehicleowners converting from gasoline to CNG could be amatter of months depending on the total number ofkilometers traveled a year (figure A6.1).

While one of the original objectives of this pro-gram was to substitute diesel with CNG, that substi-tution has not occurred because the price differencebetween diesel and CNG is not sufficient to recover

Payback for Conversion from Premium Gasolineto CNG in Argentina, 1999 Fuel Prices

F I G U R EA 6 . 1

Source: Francchia 2000.

0

5

10

15

20

25

30

35

40

5 15 25 35 45 55 65 75 85 95

Payback in months

Annual km traveled (thousands)

the incremental cost of NGVs within a reasonable pe-riod. As a result, no CNG buses are in regular opera-tion today, while diesel is actively competing withCNG to capture the taxi market from CNG.

In stark contrast to Argentina, the government ofNew Zealand was heavily involved in the NGV pro-gram from the outset. It provided generous financialincentives both for conversion and establishment ofrefueling stations. The number of CNG vehiclesdoubled every year, seriously stretching the ability ofthe industry to cope. The industry was so preoccu-pied with meeting the demand for conversion thatquality at times became a secondary priority, resultingin the perception of CNG as a second-rate fuel thatwas used only because it was much cheaper thangasoline.

When the new Labor government began to de-regulate the economy, withdrawing financial incen-tives for the CNG industry, the market essentiallydied. NGVs in early 2004 numbered some 1,500 com-pared to the peak of 110,000 (Harris 2004).

Observations from around the worldThe following preconditions are typically needed forsuccessful NGV programs:

� Natural gas should be available at low cost,which is the case if the country has large re-serves of natural gas.

� A NG distribution pipeline for other users ofNG should be in place.

� The government should establish a proper regu-latory framework to provide a fair and levelplaying field for all stakeholders.

� The government should establish adequatesafety and performance standards that aremonitored and enforced.

� The retail price of NG needs to be considerablyless than that of the liquid fuel. This makes NGsubstitution of diesel difficult, because the taxon diesel is low in most developing countries sothat making NG cheaper might even entail a di-rect subsidy. Making NG much cheaper thandiesel would be especially difficult in countriesthat import LNG.


� In the early days of a NGV program, a cham-pion coordinating activities among differentstakeholders and publicizing the benefits ofNGVs may be needed.

Worldwide experience with NGV has shown that:

� Poor conversions can lead to higher emissionsand operational problems, giving NGVs a badname. When gasoline vehicles were convertedto NG, one of the unpleasant surprises in indus-trial countries was that the converted NGVswere found to be more polluting when testedfor emissions in cases where recent model yearvehicles had been converted.

� NGVs are especially suitable for high-usage ve-hicle fleet operators who can exploit economiesof scale.

� There are consistent reports that NG busesmanufactured in the early 1990s were not onlymore expensive to purchase but were also about30 to 40 percent more expensive to maintain andhad considerably reduced reliability. Whilemany of these problems are being overcome,heavy-duty NG engine technologies are still be-ing refined to achieve performance comparableto that of their diesel equivalents.

� Successful NG bus programs are based on dedi-cated OEM, and not converted, buses. Conver-sion of existing diesel vehicles has typically notbeen acceptable to owners or users.

� Transit buses in many, if not most, developingcountries are cash strapped, partly on account offare controls. As a result, buses are not properlymaintained and operators are not in a positionto purchase more expensive NG buses, provideextensive training to all their staff on this newtechnology, and accept the possibility of morerepairs to deal with greater frequency of busbreakdowns. High emissions from diesel busesare not merely a result of the choice of fuel. Theyare symptomatic of deeper problems, and thesame problems may condemn NG bus pro-grams to failure.

� The number of refueling stations and NGVsmust be balanced so that there are no unaccept-

ably long queues for refueling on the one handand underutilization of filling stations on theother. During the early days of NGV programs,the government may consider giving incentivesto provide a critical mass of both, such as bygranting permits where filling stations were notallowed earlier (Argentina) or by providing sub-sidies of limited duration.

Liquefied Petroleum GasLiquefied petroleum gas is a mixture of light hydro-carbons, mainly propane/propene and butanes/butenes. LPG is a product of refining petroleum aswell as a component of natural gas, and a key de-mand for the product in the developing world hasbeen as a household cooking fuel. LPG is easier to dis-tribute and store than CNG; it requires pressuresranging from 4 to 13 atmospheres, compared with 200atmospheres for CNG. Among LPG’s good environ-mental features are its limited amount of highly reac-tive hydrocarbons and its relatively low sulfur con-tent. It does, however, contain olefins, which arephotochemically reactive.

Although the octane number of LPG is not as highas that of natural gas, LPG has excellent antiknockcharacteristics. Propane has an antiknock index of104, allowing dedicated propane vehicles to take ad-vantage of engines with slightly higher compressionratios than can be used with gasoline. LPG, being agas, gives similar exhaust emission reduction benefitsto those of natural gas.

As of the early 2000s, there were in excess of 4 mil-lion LPG-powered vehicles in use around the world.Italy led the world with more than 1.1 million LPG-powered vehicles, followed by Australia with over500,000, North America with over 400,000, and theNetherlands in excess of 360,000 vehicles using LPG(LPG Autofuel Systems 2002). Since then, the numberof LPG-powered vehicles has fallen in some countries,while it has risen sharply in others. By early 2004,there were 1.6 million LPG-powered vehicles in theRepublic of Korea. Three-wheel taxis in Bangkok runon LPG, as do many taxis in Japan. Most substitutefor gasoline, although there are some heavy-duty ve-


hicles running on LPG instead of diesel. It should benoted that two-stroke engine LPG vehicles, such asthe three-wheel taxis in Bangkok, provide essentiallyno benefits with respect to particulate emissions com-pared to two-stroke engine gasoline vehicles, andemit considerably more particulate matter than four-stroke engine gasoline vehicles (Hare and Carroll1993).

The main potential problems in introducing LPGto the transport sector have to do with sources of sup-ply and the distribution system. Demand for LPGworldwide is estimated by one industry analyst togrow at 3.5 percent per year between 2000 and 2010;annual petroleum demand growth was only 1.8 per-cent during the same period (Chandra and others2001). High growth of LPG demand in the Far Eastcoupled with other factors made the East of Suez re-gion a net importer of LPG for the first time in 2001.This change, along with greater price volatility seen inrecent years, has raised questions about supply stabil-ity. On the distribution side, LPG is stored under pres-sure both inside the vehicle and in the refueling tanks.Special refueling equipment is needed to transfer thepressurized liquid from storage tanks to the vehicleand to ensure that no LPG escapes during refueling.The required investments in LPG distribution and re-fueling stations have not been made in most develop-ing countries, constraining widespread use of LPG.

As with NG, LPG displacing conventional dieselwill have especially beneficial effects on particulateemissions. LPG buses are also quieter. Conversion ofdiesel engine vehicles to LPG, however, is as complexas is conversion to CNG, as the compression ratio hasto be decreased and a spark ignition system added.The city of Vienna has a fleet of more than 400 LPG-fueled buses, the largest in the world. The operatorsreport higher fuel consumption and increased servic-ing requirements, in addition to higher bus purchasecost.

Electric/HybridFor public transport vehicles direct collection of cur-rent by trams or trolleybuses is well established, withthe limitations being the inflexibility of routes and the

typically higher expense compared with diesel pro-pulsion. For private automobiles, flexibility require-ments preclude direct collection of current.

Historically, there has been a great deal of interestin battery-electric vehicles as zero tailpipe emissionvehicles. On account of lack of sufficient progress inbattery technology, however, battery-electric vehiclesare no longer considered a viable mainstream technol-ogy by most industry analysts. Recently, increasing ef-forts have been directed to hybrid electric-internalcombustion engines rather than battery-electric-en-gine vehicles. A hybrid combines an electric powersource and a fuel to drive the vehicle.

Battery-electricThe battery of an electric vehicle is central to its fuelsystem and its success. Lead-acid batteries are cur-rently used as a low-cost option in electric vehicles.Nickel-metal hydride and nickel-cadmium offer agreater driving range but are more expensive thanlead-acid batteries. Other technical options are verycostly or still under development. There is no consen-sus as to what type of battery may be best for the fu-ture. Depending on the battery type and the voltageused, battery recharging can take 4–14 hours. Becausethe driving range is short, batteries also need to be re-charged frequently. Lead-acid batteries emit hydrogenas they recharge, so indoor recharging facilities mustbe well ventilated.

The main disadvantages of electric vehicles are thelength of time needed for recharging them, theirmuch shorter driving ranges, and their higher pur-chase costs. Battery life is also limited to a number ofrecharges, and lead-acid battery life is reduced if bat-teries are fully discharged. Battery charge and dis-charge efficiency are quite poor for lead-acid batteries.The danger of electric shocks is a concern. Crashwor-thiness of battery-powered vehicles is still being stud-ied. The economics of electric vehicles depend, amongother things, on the price of electricity. The long-termviability of electric vehicles should be evaluated fromthe standpoint of market-based energy pricing. Giventhe current state of the technology, electric vehicleswould not be expected to have widespread applica-tion, but could play a useful role in reducing emis-


sions along fixed routes and can be especially attrac-tive in countries with abundant sources of low-costelectricity.

HybridHybrid vehicles are most appropriate for circum-stances in which there is a lot of stop-start driving orwhere large loads have to be carried, requiring largeinternal combustion engines (such as buses and deliv-ery trucks), because of the characteristics of hybriddrives. Some industry observers view gasoline/dieselhybrid-electric as a bridge to fuel cells. Diesel-hybridvehicles in particular offer a great potential for reduc-ing greenhouse gas (GHG) emissions.

Heavy-duty hybrid vehicles are still undergoingdevelopment and are not yet commercially competi-tive. The New York City Metropolitan TransportationAuthority tested five diesel-hybrid transit buses be-tween 1998 and 2000. The experience was positive“for a new technology,” with emission benefits match-ing those of CNG buses (the hybrid buses wereequipped with particulate filters) and fuel economysurpassing diesel or natural gas. However, hybridbuses were less reliable than diesel buses, starting offhalf as reliable, and bus availability during morningpeak service did not reach the target during the testperiod September 1998 to March 2000 (MTA NewYork City Transit Department of Buses 2000).

Given the evolving nature of this technology andtechnical issues to be addressed before hybrid ve-hicles become commercially viable, hybrids are un-likely to play a significant role in mitigating vehicularemissions in developing country cities in the nearterm. However, just as in the case of battery-electricvehicles, but perhaps more so, they can play a care-fully targeted role in special circumstances, includingas fleet vehicles, and along heavily polluted trafficcorridors.

BiofuelsThe two most commonly used biofuels in vehicles areethanol and biodiesel. Pure ethanol is used for trans-portation in Brazil, but its use has been declining inrecent years. An alternative use of ethanol is to mix it

with gasoline. Biodiesel is used in compression igni-tion engines, blended with petrodiesel or as neat (thatis, 100 percent) biodiesel.

EthanolEthanol has high octane and relatively clean combus-tion characteristics. Ethanol can be used neat orblended into gasoline. The presence of oxygen inethanol facilitates combustion, reducing CO and HCemissions from old-technology vehicles (in contrast tonewer models, which are equipped with oxygen sen-sors). Compared with diesel, ethanol produces sub-stantially less particulate emissions. The emissions offormaldehyde, which is a toxin and an ozone precur-sor, are higher from ethanol than from conventionalliquid fuels.

When the amount of ethanol in gasoline is rela-tively low—for example, 10 percent ethanol blendedwith 90 percent gasoline—ethanol increases the vola-tility of the resulting fuel mixture significantly. Pro-duction costs of specially formulated low-volatilitygasoline to compensate for the high volatility of theethanol-gasoline blend are normally higher. Ethanolhas slightly more corrosive properties than gasoline.Because of its miscibility with water and its corrosiveproperties, ethanol should be handled separatelyfrom gasoline during distribution.

The largest “biofuel” vehicle program in the worldhas historically been Brazil’s Proálcool. Ethanol hasalso been extensively used in the United States to sat-isfy the oxygen requirement under the 1990 Clean AirAct Amendments and more recently to replace me-thyl tertiary-butyl ether (MTBE). Fuel-grade ethanolin these countries is produced by fermentation ofsugar from grains (more than 90 percent of ethanolproduction in the United States comes from corn) orsugarcane (in Brazil), followed by distillation. Assuch, the price of ethanol is determined by the pricesof the starting materials (corn, sugar, wheat) on the in-ternational market, as well as the costs of extractingsugars and converting them into ethanol, which is en-ergy-intensive.

The analysis of lifecycle GHG emissions for etha-nol production is dependent on a number of factors.Assumptions about fertilizer, pesticide, and herbicide


consumption, irrigation, farm equipment, and how tohandle by-products can have a significant impact onthe final net energy balance. The calculation of netGHG emissions is also complicated by the type ofland that is used for biofuel crop production, and is-sues such as the carbon storage of different land usetypes. In the United States, the energy needed to growthe corn, harvest it, transport it, and distill it into etha-nol has been widely discussed and debated. A recentstudy conducted for the Australian Greenhouse Officeconcluded that the addition of 10 percent ethanolfrom sugarcane to gasoline has a slightly negative im-pact on overall GHG emissions in Australia (Beer andothers 2001).

Ethanol from sugarcane in the Center-South re-gion of Brazil is undoubtedly the lowest-cost ethanolin the world. Nonetheless, it is difficult to quantify theprecise production costs because of the significantsubsidies that have been provided in Brazil for invest-ment financing in the past, including for capital-inten-sive ethanol production plants. At the heart of theethanol program in Brazil is the mandatory blendingof ethanol with gasoline. Each year, a Presidential De-cree sets a range for the percentage of ethanol thatmust be used in gasoline. The actual percentage is de-termined by an inter-ministerial committee compris-ing representatives of the Ministry of Agriculture,Ministry of Finance, Ministry of Mines and Energy,and Ministry of Industrial Development and Com-merce. The blending rate has varied between 20 per-cent and 25 percent in recent years. The blending ratiotends to be increased when sugar prices are low, anddecreased when sugar prices are high. The govern-ment also bans the use of diesel-powered personal ve-hicles; requires government agencies to buy 100 per-cent ethanol-fueled vehicles; offers ethanol storagecredits to millers; and imposes an import duty onethanol (except for the intra-zone trade of ethanolwith Brazil’s Mercosur partners) to protect the domes-tic ethanol producers, 21.5 percent as of late 2003.During the same period, gasoline was taxed at R$0.57(US$0.20) per liter whereas hydrous ethanol wastaxed at R$0.05 (US$0.017) and anhydrous ethanol atR0.06 (US$0.021) per liter (USDA 2003).

In other parts of the world, the production cost ofethanol is higher. In all cases, including Brazil, promo-tion of ethanol has required implicit and explicit sub-sidies. One of the largest fiscal incentives offered isthe exemption made available to ethanol from themineral oil tax levied on gasoline of €0.65 per liter inGermany.

Compliance with extremely stringent emissionstandards requires precise control of the air-to-fuel ra-tio. Therefore, there is a tendency to seek to minimizethe range of oxygen content in gasoline. The World-Wide Fuel Charter, issued by vehicle manufacturersaround the world, limits the amount of ethanol ingasoline to 7.7 percent by volume in all categories ofgasoline, limits higher alcohols (those with three ormore carbons) to 0.1 percent by volume, does not per-mit methanol, and states that ethers are preferred toalcohols where oxygenates are used. Where up to 10percent ethanol is permitted by preexisting regula-tions, labeling at the pump is recommended (ACEAand others 2002). The European gasoline fuel specifi-cations limit ethanol in gasoline to 5 percent (see tableA3.1).

BiodieselBiodiesel is typically produced by reacting vegetableor animal fats with alcohol to produce a fuel similar todiesel. Biodiesel can be used neat or blended with pe-troleum diesel in a compression ignition engine. En-gines running on biodiesel tend to have lower hydro-carbon, particulate, and CO emissions, but higherNOx.

The technology for extracting oil from plant seedshas remained the same for the last decade or two, andis unlikely to change significantly in the future.Transesterification of oil with alcohol to makebiodiesel is a relatively simple process and offers littlescope for efficiency improvement. As a result, pro-cessing costs are unlikely to fall markedly in the com-ing years. Because plant oils and animal fats have al-ternative markets that tend to push up the feedstockprices, biomass that can grow on marginal land withlittle input and rainfall, such as Jatropha or honge


nuts, may provide an attractive alternative as a feed-stock for biodiesel for developing countries.

Similar to ethanol, interest in biodiesel as a trans-portation fuel has been driven by issues unrelated tourban air quality. The EU is a major producer ofbiodiesel where securing alternative sources of energysupply is cited as the principal reason for promotingbiodiesel. Other reasons given are the need to lowerGHG emissions and to promote economic develop-ment and maintain employment in the rural commu-nity. With respect to local air quality, the EuropeanCommission has stated that biofuels will offer little, ifany, emission advantage over gasoline and diesel inthe future (European Commission 2001). The Euro-pean Commission further acknowledges that the costof biodiesel production is two to three times that ofgasoline and diesel. Therefore, fiscal incentives in theform of large subsidies would be necessary as the keydriver for investment in liquid biofuels plants.

The World-Wide Fuel Charter limits the amount ofbiodiesel in diesel to 5 percent in its first three catego-ries of diesel, and none in the fourth (most stringent)category. Where biodiesel is blended into diesel, label-ing at the pump is recommended. The limit is moti-vated by present concerns about the effects ofbiodiesel on fuel viscosity and loss of fluidicity at lowtemperatures, corrosion, and the compatibility ofbiodiesel with seals and fuel system materials. Thebiodiesel should comply with internationally recog-nized fuel specifications, such as EN 14214 and ASTMD6751 (ACEA and others 2002).

Hydrogen and Fuel Cell TechnologyIn the long run, fuel cells have the potential to offersubstantial benefits. Fuel cells can achieve 40 to 70percent efficiency, significantly greater than the 30percent efficiency of the most efficient internal com-bustion engines, although modern diesel engines to-day can achieve maximum brake thermal efficienciesin the range of 40 to 45 percent. Fuel cells produceelectricity with high efficiency by combining hydro-gen from a fuel with oxygen from air to produce en-ergy and water. When the fuel is hydrogen itself, thereare no exhaust emissions other than water. If hydro-

gen is produced from the electrolysis of water usingelectricity generated from renewable energy, then fuelcell vehicles can be said to be utilizing completely re-newable energy sources. However, such a pathway isunlikely to be economic in the medium term, exceptin a few countries that have abundant hydroelectricor geothermal power such as Iceland. The alternativeis to form hydrogen by “reforming” hydrocarbonsfrom fossil fuels. These reformers can be on board avehicle. During reforming, CO2 and hydrocarbons areemitted as by-products.

Urban buses in particular seem suited to fuel celltechnology because of fewer technical obstacles inlarger vehicles, demand for more stringent emissionstandards because of higher population exposures tobus exhaust, and their tendency to be refueled at cen-tral refueling sites (thus avoiding the need for exten-sive refueling infrastructure). A number of fundamen-tal technical issues still remain before full-scalecommercialization of fuel cell vehicles:

� Fuel choice. Direct hydrogen is preferable fromthe point of view of the environment, but tech-nology for workable on-board storage systemshas not been developed, especially for cars. Itwill also take a long time to develop a refuelinginfrastructure for hydrogen. While distributionand storage of liquid fuels is more straightfor-ward, viable processors for re-forming liquidsinto hydrogen have not been developed yet.

� Vehicle cost. Much more work is needed tobring down the costs of fuel cell vehicles. Evenoptimistic assumptions about future technologi-cal breakthroughs suggest that it will take atleast a decade before fuel cell vehicles becomeaffordable.

� Other competitive alternatives. Fuel cell ve-hicles are competing with the moving target ofever-improving conventional combustion en-gine and hybrid-electric technologies. The busi-ness case for fuel cell cars and light trucks ismuch weaker than that of urban buses for thisreason.

Large-scale deployment of fuel cell vehicles isyears, if not decades, away. Given the amount of re-


search and development still needed before commer-cialization, it would be premature for developing

countries to devote significant resources to pilot test-ing and developing fuel cell vehicles.

139

Maintaining Vehicles: Inspection andMaintenance Programs

Lack of cultural acceptance of regular, and especiallypreventative, vehicle maintenance is one of the mostimportant contributors to air pollution from mobilesources. This is especially true in developing coun-tries. An inspection system that measures vehicleemissions to identify gross polluters and requiresthose that do not meet the standards to be repaired isstandard approach adopted worldwide to addressthis problem.

Developing countries face a number of challengesin implementing an effective inspection and mainte-nance (I/M) program. What is needed are reliabledata on active vehicle population, suitable test proto-cols, and administrative control.

Data on Vehicle PopulationMost developing countries do not have a reliable da-tabase on their active vehicle population. While newvehicles may be registered, vehicle retirement is oftennot recorded, making it difficult to identify which ve-hicles are still being driven. For example, there is noannual vehicle registration in India: vehicles are regis-tered when new and are supposed to reregister onlyafter 15 years’ service. This makes it very difficult forthe government to obtain accurate information on thevehicle population and to conduct an effective inspec-tion program: the authorities cannot be sure whichvehicles are still being operated and hence cannot es-timate the percentage of vehicles that evade inspec-tion. The problem is compounded by the fact that thedata are least reliable on the very vehicles that arelikely to be gross polluters, namely, old vehicles. Insome developing countries, it is not rare for more thanone vehicle to have the same registration number.One of the first steps in making I/M effective is to cre-

ate a need on the part of vehicle owners to comply byreporting to test centers, but this is difficult withoutan up-to-date and accurate vehicle registration sys-tem.

Test ProceduresMuch has been learned in industrial countries as towhich test protocols are suitable for I/M. Test proto-cols that result in large variations and differencesacross different test centers are not only unhelpful forthe purpose of identifying high emitters, but also de-crease the credibility of the I/M system and reducepublic acceptance. It is difficult to sustain an emis-sions inspection program when substantial measure-ment differences exist among testing stations. Theelimination of measurement differences must there-fore be a high priority. This often requires changes inthe equipment and test protocols used in the pro-grams together with calibration audits and other ac-tivities.

Dilution of exhaust gasIn I/M, concentrations of various pollutants in the ex-haust gas are typically measured. These concentra-tions can be easily lowered by entraining clean air.Therefore, monitoring the dilution of the exhaust gasmust be one of the integral elements of any test proto-col. Otherwise, it would be too easy to pass by par-tially withdrawing the sample probe from the exhaustpipe to allow sample dilution to reduce the gas con-centrations. Controlling for this in turn requires O2

and CO2 measurements in addition to the pollutantsof concern. The establishment of dilution thresholdvalues, outside of which the test is automaticallyaborted, is also required. Frequently, O2 and CO2 are

A N N E X7


not measured and their measurements are not re-quired by the test protocol.

Preventing “late-and-lean” tuning in gasolineenginesIt is easy to reduce the CO and HC emissions from anolder vehicle by delaying the ignition timing andmaking the air–fuel mixture lean. This “late-and-lean” approach reduces CO and HC, but it also re-duces engine power and can increase NOx. Withouttesting vehicles under load or measuring NO, it is notpossible to detect if the engine has been tuned “lateand lean” just to pass the test. Vehicles that have beentuned in this way to pass the test are usually retunedimmediately afterward to regain this lost power. Thispractice allows the “clean-for-a-day” brigade to tunethe engine temporarily just to pass without funda-mentally lowering the vehicle’s emission levels,thereby having no positive impact on air quality. Test-ing under load and measuring NO requires a dyna-mometer. Most developing country cities do not havedynamometer-based testing, leaving the door openfor the loophole of tuning “late and lean.”

Remote sensingRemote-sensing technology for vehicle emissions is atool for testing a large number of vehicles rapidly un-der potentially realistic conditions. Remote sensingutilizes the principles of infrared spectroscopy tomeasure concentrations of HC, CO, CO2, and NOx inthe exhaust plume of a vehicle while it is being drivenon a street or highway. Recently the ability to measuresmoke has been added. The speed and acceleration ofthe passing vehicle can be recorded simultaneouslywith an image of the license plate, making it possibleto identify vehicles and determine the conditions un-der which the measurement was taken. Existing sys-tems can measure more than 4,000 cars per hour on acontinuous basis, potentially providing a powerfultool for characterizing the emissions from the on-roadvehicle fleet. Remote sensing has been used as a com-ponent of I/M programs in North America, either toidentify high-emitting vehicles and call them back forrepair or to identify clean vehicles and exempt themfrom regularly scheduled measurement at I/M sta-

tions. It has also been used to post electronic signs in-forming passing drivers of the status of their vehicles’emissions status in an effort to raise public awareness.

Existing remote-sensing systems have a number oftechnical limitations:

� Smoke has been added only recently. Untilsmoke was added, remote sensing could beused only for gasoline vehicles.

� Measurements are limited to a single lane oftraffic.

� Heavy-duty vehicles may have exhaust loca-tions that make them inaccessible to a remote-sensing device set up for passenger cars.

The reason for the single-lane limitation is that thebeam traversing the road at exhaust height must seethe emissions from one vehicle at a time for uniqueidentification. This condition can be met by placingthe remote-sensing device on highway entry and exitramps. However, such ramps may not see all vehicles,or a representative sample. Furthermore, like exhaustmeasurements at idle, emissions of vehicles on entryand exit ramps do not show the best correlation withemissions over a whole trip. To channel part of thetraffic on heavily traveled roadways to a special mea-surement lane is not generally considered a viable op-tion. Another mode of operating a remote-sensing de-vice as an I/M tool would be to ask drivers to drivepast an off-road station at a given speed instead ofstopping to complete exhaust measurements at idle oron a chassis dynamometer.

At present, remote sensing can therefore be con-sidered proven technology only for light-duty gaso-line vehicles. For heavy-duty gasoline vehicles or die-sel vehicles, promising advances have occurred inremote-sensing research and development, but theiroperational effectiveness needs to be demonstrated infield trials. Effective remote sensing must cause ve-hicles to behave in some reasonable representativemanner, such as accelerating lightly or climbing a hillat some speed. Emissions vary substantially with loadand transient behavior, so that a careful design isneeded to limit variations in the results to an accept-able level.

ANNEX 7. MAINTAINING VEHICLES: INSPECTION AND MAINTENANCE PROGRAMS 141

Identifying high particulate emittersIdeally, diesel vehicles that emit disproportionatelyhigh levels of fine particulate matter should be identi-fied and required to be repaired. Identifying thosediesel vehicles that are high particulate emitters, how-ever, is problematic, because the test procedures cur-rently used in the emissions inspection test centers donot readily identify mass particulate emissions.

Smoke tests

The most common procedure for testing emissionsfrom in-use diesel vehicles is the “snap” (also called“free”) acceleration test defined according to the Soci-ety of Automotive Engineers’ (SAE) J1667 (in NorthAmerica) or ECE R24 (in Europe) standards. The testspecifies that, with the transmission in neutral, thethrottle pedal should be pushed rapidly but notabruptly to its full-throttle position, accelerating theengine from low idle to its maximum governedspeed. This is repeated several times and the averageof the maximum exhaust gas opacity in each test iscomputed.

Reproducibility is poor when this test procedure isutilized by a large number of testers using differentequipment. In July 2002 in Mexico, a test was con-ducted in which simultaneous free-acceleration testswere conducted using the ECE R24 procedure on fourmakes of opacity meter with different operators. Allwere taking simultaneous samples from a MercedesBenz L1217. On any particular run, the difference be-tween the lowest and highest reading was as much as700 percent. Even the two closest readings differed by30 percent.

Slight differences in the time taken to acceleratethe engine from low idle to maximum governedspeed can lead to very different exhaust opacity read-ings. Therefore, the rate of acceleration for each en-gine type needs to be precisely defined. Each instanta-neous reading should be corrected for gastemperature, pressure, humidity, and altitude, as re-quired in the SAE J1667 standard. Any dilution of theexhaust gases with clean air will lower smoke read-ings. Furthermore, it is difficult to get revolutions perminute (rpm) readings from diesel engines, particu-

larly the older ones, yet accurate rpm readings are es-sential to add to the controls as well. Finally, the ECER24 procedure, but not SAE J1667, verifies that the en-gine cylinders’ combustion chambers have reachedtheir normal operating conditions by detecting de-creasing opacity readings in consecutive tests and re-ports results only after stable conditions have beenmet.

Carefully defined test protocols need to be fol-lowed strictly to have acceptable reproducibilityacross different operators and instruments at differenttest centers. Otherwise, a conscientious vehicle fleetowner regularly checking emissions by conducting in-house smoke tests may find that properly maintainedvehicles routinely fail in the mandatory emissionstests. This and the resulting harassment of driverswere among the complaints voiced in the assessmentof the emissions program in India (Rogers 2002). Fur-ther procedural improvements can be introduced toreduce the opposite kind of error, by which grosslyemitting vehicles are allowed to pass the test.

Shortcomings of smoke tests

Even when properly administered, smoke tests stillhave two distinct problems. The first concerns thesnap acceleration test and can be addressed by usinga more reproducible (and more expensive) form oftesting, called a dynamometer test, which can alsosimulate real-world driving conditions better thansnap acceleration. The second problem cannot be ad-dressed, as it underscores the fundamental shortcom-ing of measuring smoke.

The snap acceleration test is not representative ofnormal operating conditions. More specifically, it iseasier to prepare a vehicle to pass a snap accelerationtest than a test under load using a dynamometer, es-pecially if a transient (as opposed to steady-state)loaded test is used. In Hong Kong, for example, envi-ronment officials found that diesel vehicle ownerstemporarily adjusted the fuel injection pump, en-abling high smokers to pass the snap accelerationsmoke test. The officials closed this loophole by intro-ducing the so-called lug-down dynamometer test (testconducted at full throttle, with the dynamometer load


gradually increased to slow down the engine speed sothat the engine is laboring, or “lugging”). Immedi-ately after this change, the pass rate fell dramatically(Mok 2001).

Furthermore, smoke measurements dependheavily on the driving cycle for a given vehicle-fuelcombination. One study (NEPC 2001a) shows thatsmoke opacity measurements using different drivingcycles on a loaded dynamometer were poorly corre-lated with each other for light-duty vehicles. The low-est correlation coefficient was –0.16, the highest 0.73.The correlation coefficients for heavy-duty vehicleswere higher, with the lowest coefficient being 0.20 andthe highest 0.92. This implies that the same vehiclemay have quite low readings under one driving cycleand high ones on another, and yet there may not be a“typical” driving cycle for all vehicles.

A smoke test procedure, however well carried out,cannot be used for controlling anything other thanvisible smoke. An important question is then whethervisible smoke can act as a proxy for particulate matter,the pollutant of most concern from a health perspec-tive.

Correlation between smoke opacity and mass particulateemissions

If a diesel vehicle testing program does not lead to thereduction of fine particulate emissions, then it hasfailed in its ultimate objective, namely, the reductionof pollutants that damage the public’s health. Thequestion immediately arises, therefore, as to howclosely correlated “smoke” is with fine particulatematter.

To answer this and other questions, the NationalEnvironment Protection Council (NEPC) of Australiacommissioned eight projects to collect data on the die-sel vehicle fleet and emissions characteristics with theobjective of developing cost-effective emissions man-agement measures. (The Environment Protection andHeritage Council Web site, included in the References,provides information about the projects.)

Figure A7.1, taken from a project report (NEPC2000), indicates a very poor correlation between vis-ible smoke measured using SAE J1667 snap accelera-tion and mass particulate emissions measured in a

dynamometer test using a driving cycle for estimating“real-world” emissions from vehicles in urban areas,called the Composite Urban Emissions Drive Cycle(CUEDC). CUEDC consists of four segments: con-gested, minor roads, arterial roads, and highway/freeway driving. The figure illustrates that a numberof high particulate emitters have quite low scores onsmoke emissions registered during the free accelera-tion test, while some of the high “smokers” have rela-tively low particulate emissions compared to the truegross polluters. That is to say, snap accelerationsmoke tests run the danger of classifying gross pollut-ers as relatively clean and of classifying low pollutersas high emitters.

The report also compares the results of the so-called 10-second smoke test—used in enforcingAustralia’s national standard for smoke emissions—to smoke opacity and mass particulate emissionsmeasured in transient dynamometer tests. In a testspecifically done on an incline, 32 percent of thosethat were identified as smoky vehicles were alsofound to be high particulate emitters in the CUEDC,but 68 percent of smoky vehicles were found to below particulate emitters. Overall, the on-road smokechecks classified a higher number of vehicles as highemitters than mass particulate measurements.

Arising from this, an unfortunate scenario, fromthe point of view of managing a vehicle inspectionand maintenance program, would be one in which

Correlation between Visible Smoke and MassParticulate Emissions

F I G U R EA 7. 1

Source: NEPC 2000.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 20 40 60 80 100

Maximum % opacity in snap acceleration

g PM/km in CUEDC


vehicle repair lowers particulate emissions but in-creases smoke. Such a case was indeed confirmed inthe NEPC’s diesel program whereby for two vehiclecategories (1996–2000 diesel buses heavier than 5 tonsand goods vehicles between 12 and 15 tons for thefirst, and 1996–2000 goods vehicles heavier than 25tons for the second), particulate emissions decreasedby 38 percent and 14 percent, respectively, on averageafter repair, but opacity readings increased by 29 per-cent and 10 percent, respectively (NEPC 2001b).

In light of the above, snap acceleration smoke testscannot be viewed as a means of identifying high par-ticulate emitters, but rather as diagnostic tests to iden-tify malfunctioning and defects among older enginevehicles with mechanically controlled fuel systems.For this category of vehicles, these smoke tests may beespecially helpful for identifying tampering to in-crease power by overfueling. However, given thepoor correlation between smoke and particulate emis-sions, and the poor correlation among the results ofsmoke tests on the same vehicle using different driv-ing cycles, it would make sense to set relatively le-nient standards to identify the most serious smokeemitters so as to minimize chances of false failures.

Snap acceleration smoke tests are also ineffectivefor modern, electronically controlled engines or turbo-charged engines with boost control. For these vehiclecategories, an alternative test procedure is required.Studies to date suggest that at a minimum dynamom-eter-based loaded tests are needed, but they are ex-pensive to set up for heavy-duty diesel vehicles. Theseries of studies in Australia recommend a short dy-namometer-based test with transient acceleration seg-ments using a laser light scattering photometer tomeasure mass particulate emissions. However, thistest is still at the pilot stage.

Finally, it is important to view the limitations ofsmoke tests in broader perspective. Smoke is a publicnuisance and harms public health. High smoke emis-sions, even if particulate emissions prove to be rela-tively low, suggest that there is something wrong withthe vehicle settings or parts. Equally importantly, thefact of having to report for emissions testing willprompt some vehicle owners to pay closer attention

to vehicle maintenance and exhaust emissions. Giventhe technical problems associated with measuringparticulate emissions in a garage setting, smoke testswill therefore continue to play an important role inemissions testing programs for the foreseeable future.

Approaches to monitoring diesel vehicle emissions

Identifying gross diesel polluters is significantly moredifficult than identifying gross gasoline polluters. Oneserious problem is the lack of a relatively inexpensiveand quick method for measuring particulate emis-sions, the pollutant of most concern in the majority ofdeveloping country cities. There is no simple roadmap, and a multi-thrust strategy is needed:

� In order to encourage the development of ad-equately equipped and staffed service and re-pair facilities, more efforts should be directed atimproving vehicle emissions inspection and en-forcing standards.

� There is a need to raise public awareness andcreate market conditions that discourage over-loading of commercial vehicles.

� There needs to be tighter control over the qual-ity of spare parts.

� The most commonly used test for diesel ve-hicles—the snap acceleration smoke test—canplay a limited but useful role, provided that thetest procedure is carefully defined to ensure re-peatability and the test protocol is strictly en-forced.

� The smoke standards for snap acceleration testsshould be lenient; if not there is a serious dangerof having a high false failure rate. The testshould be seen primarily as a means of identify-ing the worst gross polluters among old technol-ogy engine vehicles.

� For modern engine vehicles as well as high-an-nual-km commercial vehicles, serious consider-ation should be given to adopting dynamom-eter-based tests. This will also enable settingtighter smoke standards.

� Given the high costs of dynamometer testing fa-cilities, especially for heavy-duty diesel vehicles,using on-road visual smoke checks as a screen-


ing tool—whereby those judged to be visiblesmokers are sent to test centers with dynamom-eters—may help reduce the overall cost of thediesel emissions inspection program. While vi-sual checks are subjective and potentially moreprone to corruption, using them as a screeningtool so that those who conduct the checks can-not fine motorists may help to minimize corrup-tion.

� Inspecting underhood equipment, such as theair filter, or measuring manifold boost may be arelatively inexpensive way of identifying highparticulate emissions.

Administrative ControlEven when all test centers are in the hands of the pri-vate sector, local authorities must dedicate significantresources, personnel, and effort in supervising andcontrolling I/M programs. More details are given in arecent publication (ESMAP 2004).

SupervisionIn an environment where the requirement to possesscurrent certificates is enforced, the black- marketvalue of the certificate increases. Certificates andstickers should be issued to the inspection centers ona controlled basis and their use supervised. Both cer-tificates and stickers should ideally incorporateantiforgery design elements similar to those used inbanknote production. The stickers should incorporatea highly visual design and be easily visible at a dis-tance of five meters, enabling enforcement personnelto determine whether certification is current or not.

To be effective, the test protocol must minimizethe impact the test technician can have on the test’soutcome. It is good policy not to make the test resultsavailable to the test technician in the test lane until thevehicle owner has been officially informed of the finaldecision (pass or fail). Otherwise, it is common fortesters to prevent rejects from occurring by tamperingwith the lane computer, the test procedure, or the ve-hicle. The availability of real-time emissions measure-ments even helps the tester to generate a false pass forvehicles that would otherwise have failed the test.

The computer should take the gas reading after a pre-established time. As long as this decision is left to thetester, there can be tampering until a pass result isgenerated.

Vehicle owners must sense a strong need to com-ply with I/M; that is, they must feel compelled to re-port to testing stations. This requires that stickers becontrolled by the government and be difficult to fal-sify, and that the sticker have a highly visual designthat enables any police officer to easily identify if thevehicle has a current certificate. It also requires thatthere be enough traffic police or their equivalent em-powered to stop and fine vehicles without currentstickers.

Centralized software developmentDeveloping one software package to be used by all in-spection centers merits serious consideration. Thatcomputer software package should contemplate en-suring compliance with test protocols, proper opera-tion and maintenance of equipment, security, datacapture and analysis, and administrative functions.

Test protocol control and improvements

� Better consistency and repeatability in the test

results and control of false passes. Running thetest protocol under computer control allowsmany of the test variables to be dynamicallyverified under real-time second-by-second con-ditions and allows multiple readings to be ob-tained under precise repeatable conditions toeliminate instantaneous inconsistencies. The re-corded second-by-second data are useful for sta-tistical analyses designed to detect fraudulentpractices during the test. In the snap accelera-tion test to measure smoke opacity in diesel ex-haust, for example, the adoption of computercontrol of the test procedure allows many other-wise-uncontrolled variables to be managed: thesoftware is able to determine the characteristicsof the diesel engine, particularly its low-idlespeed, its rated maximum power speed, and thetime it should take to accelerate between thetwo, if these have been included in a master ref-erence table within the program. With this infor-


mation the software is able to validate if the testhas been correctly performed. Without com-puter control the technician is able to start thetest at a higher-than-low-idle speed, end the testat a lower-than-maximum speed, and accelerateat a slower rate—all of which reduce the smokeemissions readings.

� Dynamic validation of all the test protocol

constraints. These include exhaust gas dilution,engine or vehicle speed, engine load, and otherfactors. Every second the computer performschecks to see if these parameters are within lim-its, that the test equipment is not in a low-flowcondition, and if the equipment has been tam-pered with. Should any of these parameters befound to be out of bounds, messages are givento the tester (and logged in the computer data-base) to correct this situation and require thatthe test be restarted.

� Differentiation of emission standards. Reliableand consistent vehicle identification and testdata in centralized databases are essential forperforming the statistical analyses required todefine the emission limits. These can be ob-tained only when the test process is computercontrolled. Different vehicles may require differ-ent emission standards depending on their de-sign and size, and this can be accommodatedonly when the vehicle can be reliably and con-sistently identified through a database. An ex-ample of this is oxygen. For most gasoline andgas-fueled vehicles a high oxygen reading in theexhaust shows that the vehicle is not in an ad-equate mechanical condition (very lean air–fuelmixture or leaks in the exhaust pipe) or that thetest has not been performed correctly. Howevera few vehicles equipped with three-way cata-lysts have air injection into the center of thecatalytic converter to promote the oxidation pro-cess, and these have higher oxygen readings inthe exhaust gas. The use of a computer-con-trolled test allows adequate oxygen limits to beapplied to both types of vehicle.

Proper equipment operation

� Automatic computer-controlled random cali-

bration audits. These audits are needed to en-sure that the instruments are within their correctspecifications.

� Automatic equipment calibration. The soft-ware can improve the measurement accuracy byautomatically calibrating the equipment at inter-vals that depend on each specific instrument’sproven stability and usage pattern. Intensivelyused I/M gas and opacity meters need to becalibrated with reference gases and filters atleast once per day when the equipment isstarted up and should be referenced to zero be-tween every test. Computer- controlled testingallows the instrument’s calibration to be verifiedand corrected at predefined and modifiable pe-riods. For example, an optical gas analyzershould be routinely auto-calibrated with refer-ence gases every day before testing commences.On days when the work load is particularly in-tensive a more frequent re-calibration is highlybeneficial. Re-calibrations should also be consid-ered after high emitters have been tested sincethere is a tendency to saturate the measuring cir-cuits, which can cause the next vehicle tested tofail. The calibration history of each instrumentcan be used to determine the frequency of cali-bration it requires. A new optical bench thatgenerates stable and consistent readings can becalibrated less frequently (with considerablesavings in reference gases and time) than anolder, less stable instrument.

Data capture and analysis

� Correct identification of vehicle. The identifica-tion of the specific vehicle and its owner can beimproved by using computerized data entrysoftware that automatically consults lookuptables to verify that the information is correctlyentered and the correct test procedure is ap-plied. Great care needs to be taken in the dataentry to ensure that the data are entered every


time without errors. Centralized databases al-low the software to check for consistency andcorrect errors at source. This ensures that thecorrect test procedure, limits, and follow-up ac-tions are chosen for that vehicle. Only wheneach vehicle is correctly and consistently identi-fied can different test protocols be reliably ap-plied to different vehicle types with correspond-ing differentiated emission standards. Thealternative of applying the same standards to allvehicle types is suboptimal because the stan-dards are likely to be too lenient for those withnewer technology and yet very stringent forother vehicle types.

� Use of bar-code scanners to improve data entry

accuracy. When the centralized databases con-tain tests for a specific vehicle, registering theprevious test certificate’s barcode with a scannerallows the captured and validated data for thatvehicle and its owner to be accessed. Thus thetest technician only has to enter and validateany changes that might have taken place againstother official documents. These include changesof ownership for that vehicle and changes of ad-dress for the owner.

� Real-time electronic transmission of data. Au-tomatic real-time electronic transmission of thevehicle information, test data, and certificationstatus to a central host computer system servesto minimize tampering of captured data and en-ables stringent automatic quality assurance pro-cedures with auditing of test results and inspec-tion equipment status by the central hostcomputer system. Statistical analyses and othertechniques need to be incorporated into the soft-ware on the central host computer system thatcontinuously monitors the test results, calibra-tion results, and maintenance requirements ofeach test center and each tester-lane combina-tion. Common make/model and model-year ve-hicles, for example, should generate similaremissions distributions in all test centers andwith all test technicians. If one center is found tohave emissions results for that particular type of

vehicle that are substantially lower than thoseregistered in the other test centers, it is probablethat it is using some technique to artificiallymodify the test results. Such findings, if con-firmed, should lead to the temporary or perma-nent suspension of that center or stricter vigi-lance by the controlling authority.

Security measures

� Improved equipment security measures and

electronic detection of tampering attempts. Ifthe test technician or center operator were al-lowed to modify the characteristics of the mea-suring instruments (which are partly defined ineach instrument’s computer code) or the charac-teristics of the test equipment or the analog-to-digital conversion of the measurements made ormany other conditions, false readings can be ob-tained that would help a polluting vehicle ob-tain a pass certificate. Some of the measuresused to prevent such modifications include:

– Individual password codes that allow onlyauthorized individuals to access specific areasaccording to governmentally defined guide-lines. (For example, a test technician shouldnot have access to maintenance functions andmaintenance personnel should not be allowedto perform official emissions tests.)

– Electronic locks on the equipment cabinetsthat can be opened only when an activatedand authorized password code is entered.

– Automatic diagnostic checks every time acabinet is closed to ensure that unauthorizedmodifications have not been performed.

– Electronic entry and exit logs transmitted tothe central server of all activity that couldhave resulted in undue adjustments.

� Automatic printing of digitally signed and

validated certificates. As controls becomestricter it will become increasingly difficult toobtain a pass-certificate fraudulently and un-scrupulous center owners will look for moretechnical and sophisticated means of supplyingtheir clientele with fraudulent certificates. Ele-


ments such as digital signatures then need to beincluded in the certificates to identify forgeriesand to identify fraudulently generated certifi-cates. A digital signature is a set of charactersthat have been generated by high security codethat takes into account the contents of the certifi-cate in such a way that if any of the certificate’scontents were modified, they would no longermatch the signature and the modification can beeasily detected. Thus no two certificates couldhave identical digital signatures. These are usedin conjunction with digital fingerprints that offera synopsis of the signature in a form that is easyto enter into the computer.1 This allows the com-puter to validate, for example, when a vehicle ispresented for testing that its current (or previ-ous) certificate has not been fraudulently gener-ated or modified.

� Improved data and certificate security mea-

sures to control electronic tampering. The elec-tronic information also needs to be protected.This can require encryption (where the informa-tion cannot be read or deciphered without a spe-cific key),2 per-register and per-table3 digital sig-

natures and fingerprints (where the informationcan be read, but not modified without detec-tion), and protected operating systems and data-base structures.

Administrative measures

� Remote lock-out. Remote lock-out of test andinspection equipment when inconsistencies ormalpractice are found should be the first line ofdefense against fraudulent practices at test-cen-ters.

� Accounting. There should be automatic certifi-cate accounting and control procedures fromgovernment to test centers and to end usersverified against tests performed and results ob-tained. An onerous part of the controlling au-thorities’ test-center supervisory process is keep-ing track of the test certificates issued to eachcenter, and of those, which were issued to endusers, which were returned to the authority fortechnical or other reasons, and which have notbeen accounted for. The software needs to vali-date the certificates’ reported use against thecentralized test databases to ensure that certifi-cates are not just printed out with a spreadsheetprogram to issue false passes to dirty vehicles,bypassing the technical emissions test alto-gether.

� Checking compliance. Vehicular enforcementprocedures need to be strengthened to ensurethat all vehicles do satisfactorily complete theirdesignated inspection and certification require-ments within the allotted time period. Once acentralized and comprehensive database hasbeen established, information can be generatedby the computer system on those vehicles thathave not turned up for testing within their allot-ted time period or have not returned for re-testafter having obtained a fail certificate. This in-formation greatly helps the vehicle enforcementprocess.

Auditing

It is difficult for a salaried on-site inspector to ad-equately control an authorized testing station that is

1A digital signature, a long chain of characters, uniquely de-fines the certificate. A fingerprint which is shorter for ease of en-tering into the computer is unlikely to be unique. For example, ifone has to type it in, a string of 6 characters is a good fingerprint.If one uses only capital letters, the number of options in this stringis 26 to the power of 6 which is unlikely to give a unique combi-nation: 266 = 308,915,776 and a key that has only 300 million com-binations is cryptographically speaking very weak and couldprobably be broken using a normal desktop computer in less than24 hours. A signature (which does not need to be typed in) cancontain more options per character and the string could be 254characters long, which gives a unique combination. If the digitalfingerprint is put on the certificate in the form of a barcode, a fargreater number of combinations can be contemplated.

2There are many ways of encrypting data. In all cases, oneuses a formula or a specific key (character string) to convert theplain-text message into its encrypted form. One then needs to usethe same or another formula or key to convert the encrypted mes-sage back into its plain text form. Without the formula or charac-ter-string in one’s possession the encrypted message cannot beconverted into a readable format.

3Two things need protection in the database: (1) data in a spe-cific test record should not modified (for example, ensure that theemissions readings that were reported by the test equipment havenot been manually changed in the database); and (2) that recordsin the database should not deleted and that new records shouldnot added, bypassing the test equipment. This requires protectingthe contents of each record (per-register) and ensuring thatrecords are not added or deleted from the database (per-table).


making a lot of money fraudulently. I/M programsshould make use of remote auditing to distance theirstaff from the temptation of turning a blind eye tofraudulent practices in exchange for monetary pay-ments. The local authority needs to invest in remote,computer-based auditing of all centers. The nationalgovernment could develop such computer programscost-effectively, since the same requirement will existin every city that adopts such a system.

Electronic data transmission is essential. Unlessfresh data from all tests performed are readily avail-able, it will be very difficult to supervise and audittest operations effectively.

Calibration audits play a very important role inensuring that the test equipment is correctly main-tained and eliminating the perennial problem of thesame vehicle producing different test results in differ-ent centers. Gas, opacity-filter, and dynamometer cali-bration audits should be performed by independentaccredited materials standard laboratories on eachtest lane regularly. The gases used should be traceableto international standards and certified in accordancewith the U.S. EPA Protocol G1 or G2 or other interna-tionally recognized protocols. Similarly, a set of inter-nationally traceable neutral filters should be used toevaluate the linearity of the opacity meters.

Experience in Mexico CityThe I/M program in Mexico City is widely acknowl-edged to be one of the most successful in developingcountries. The Mexico City government initiated anI/M program in 1982 as a voluntary exercise. Theprogram has undergone a number of changes since,adopting more reliable and stricter testing proceduresin order to reduce the number of vehicles obtainingfalse passes. An independent assessment of the pro-gram, including a detailed analysis of emissions data,was carried out in 1999–2000 (ESMAP 2001a). Morerecently, remote sensing was used in 2003 to collect in-dependent data on the emission levels of vehicles asthey are driven on the road. Data analysis is currentlyunderway.

The annual inspections, made mandatory in 1988for certain age vehicles, were initially conducted in

the test-only centers operated by the city government,but soon afterwards independent test-and-repair ga-rages were authorized. In 1991, a proposal was madeto create independent, multilane, test-only “macro-centers” in which some of the lanes would beequipped with dynamometers (enabling dynamicloaded-mode testing). By 1993, 500 test-and-repaircenters and 24 macrocenters, all privately owned,were in operation. This side-by-side operation al-lowed a direct comparison to be made between thetest-and-repair centers and the test-onlymacrocenters.

The test-and-repair centers were convenient forvehicle owners in that they provided a one-stop solu-tion and eliminated the “ping-pong” effect of the ve-hicle owner being caught between a garage that ar-gued that it had correctly repaired and tuned thevehicle and a macrocenter that pronounced the ve-hicle to be in noncompliance. As a result, most privatevehicles went to the test-and-repair garages, whereasall vehicles that were not privately owned had to goto the macrocenters for the dynamometer test, whichwas unavailable at the test-and-repair centers.

From the point of view of quality control, the test-only macrocenters were far easier for the governmentinspectors to supervise. They also allowed better tech-nical and administrative control to be enforced. Theownership of these centers was concentrated in a fewindustrial groups specializing in emissions inspection,facilitating the adoption of new technology and gen-erating more uniform results among centers.

Over time, the quality of testing from the test-and-repair centers degenerated. There was surplus capac-ity of test centers, increasing the incentive to cheat.The garages soon found that they could make moremoney by cutting back on the cost of the repair ser-vices performed if they cheated on the emissions test-ing. The situation deteriorated to the extent that an es-timated 50 percent of the vehicles that went throughthe test-and-repair centers eventually obtained theirapproval certificates fraudulently. Public opinion wasthat it was a highly faulted emissions control pro-gram, and it was very close to being shut down per-manently.


These problems led to the program being com-pletely restructured in 1995. Despite political opposi-tion, the licenses were withdrawn from all 600 test-and-repair centers, while the number of test-onlymacrocenters was increased. A series of stringentquality assurance controls and technical changes wereadded to the multilane center operation and a newpublic identity was generated, repositioning them astest-only “verificenters.”

In addition to making some technical adjustmentsto the testing procedures, the verificenters introducedelaborate precautions to prevent individual testersgiving false passes. These included the use of “blind”test lanes where the tester did not see the results ofthe test, which were available only at the exit from thestation; central computer and video monitoring oftesting; and technical audits of centers by governmentinspectors. Because of these actions, the proportion offailing tests increased several-fold.

In spite of these quality control measures, the sys-tem had serious shortcomings. It was estimated thatduring the first semester of 1997, 19 percent of ve-hicles obtained their certificate because of a loopholein the standards and test procedures. Garages wouldtune vehicles “late and lean” or disconnect air hosesfrom the inlet manifold. Once the test had beenpassed, the vehicle would be retuned. These tech-niques sometimes reduced the engine power duringtesting to an undriveable level and could also have in-creased NO emissions. However, they effectively, buttemporarily, reduced HC and CO emissions andcould not be detected by the test procedures in place.

To address technical problems of the testing proce-dures, considerable work was done by the MexicoCity government during 1995–96 to define a new pro-tocol—acceleration simulation mode—from which ahybrid version went into effect for the second semes-ter of 1997. The objectives of modifications in the testprotocol were to generate more reproducible test re-sults, reduce measurement uncertainties, permit theuse of stricter test limits, and reduce false approvals.By 2001, a fine of up to US$40,000 was imposed ontest centers located in the Federal District of MexicoCity caught not following the test or administrative

procedures. During the first semester of 2000, NOlimits were established, eliminating the loophole inthe I/M program whereby owners of polluting ve-hicles could still pass by tuning “late and lean.”

The experience in Mexico City shows that for anI/M program to be effective, several conditions haveto be met:

� The government must be willing and able to in-vest in the resources, staff, and effort in auditingand supervising the program to guarantee itsobjectivity and transparency. This includes re-mote-auditing and the development of central-ized software that controls not only data entry,storage, and analysis, but also equipment cali-bration, zero-referencing, and test protocols.

� A legal framework has to be established that al-lows sanctions to be applied for failure to carryout the testing protocols correctly. The testingstations must be subject to monitoring by inde-pendent bodies, and in cases of noncompliance,sanctions must be applied.

� Testing protocols should be designed to mini-mize the chances of individual testers givingfalse passes.

� The certificate for passing the test has to be easyto monitor, and there should be sufficient moni-tors (for example, traffic police) to ensure a highprobability of catching vehicles that do not dis-play such a certificate.

� The fine for not displaying or not having a legalemissions test certificate must be high enough toact as an incentive to pass the test.

� The testing technology has to be able to preventthe use of temporary “tuning” that enables a ve-hicle to pass the test but cannot be sustained forregular driving. In the absence of such a tech-nology, motorists and garages become adept atcircumventing the purpose of the testing proce-dure—to identify high-polluting vehicles.

� All testing centers must be subject to equally rig-orous implementation of protocols and inspec-tion of their procedures; otherwise, owners ofthe highest-polluting vehicles easily identify the“softest” centers for passing the test.


� The optimal number of centers relative to thevolume of traffic to be tested has to be li-censed. If there are too many small centers,

the rigor of the tests tends to be watereddown as each garage tries to increase its mar-ket share.

151

Estimating the Health Impacts ofAir Pollution

Air pollution has been associated with a variety of ad-verse health effects (see table A8.1). These include im-pairments in lung function, increased incidence ofchronic bronchitis, exacerbation of chronic respiratorydisease (that is, asthma) or coronary disease (such asangina), and premature mortality from respiratoryand cardiovascular disease. Less serious effects in-clude increased incidence of acute respiratory illness(such as colds and sinus problems) and subclinical ef-fects (such as itchy, watery eyes).

Selecting the Health Effects to Be StudiedThe most important health effects, in terms of eco-nomic damages that can be assigned monetary val-ues, are premature mortality and increased incidenceof chronic heart and lung disease. The air pollutantsthat have shown the strongest association with pre-mature mortality and heart and lung disease are PMand airborne lead. Air pollutants have been associ-ated with hospital admissions, respiratory infections,asthma attacks, restricted activity days (RADs)—dayson which a person cuts back on his or her normal ac-tivities, but does not necessarily miss work or stay inbed—and days of work loss. SO2 and NOx do nothave such significant direct effects, although they dohave important health consequences because of sec-ondary particulate formation: sulfates and nitrates re-act with ammonia and other substances in the atmo-sphere to form particulate matter, such as ammoniumsulfate and ammonium nitrate.

How Are Health Effects Estimated?Estimating the health impacts of air pollution reduc-tions (World Bank 2003a) entails three steps. First, thedemographic groups susceptible to air pollution and

associated health outcomes are identified based al-most exclusively on epidemiological studies. Thesestudies determine relationships—referred to as con-centration-response (CR) functions—between air pol-lution and health effects in human populations. CRfunctions empirically explain variations in the num-ber of cases of illness or death observed in a popula-tion based on changes in the ambient concentrationsof the air pollutants and other known explanatory fac-tors. These other factors, called confounding factors

(those that also affect health outcomes, making it diffi-cult to attribute cause), include demographics (suchas age, gender, marital status, diet, body mass, smok-ing, health habits, occupational exposure, education,and income); other pollutants; and time-varying fac-tors (temperature, seasonality, and day of week). CRfunctions may apply to the whole population or tospecific demographic groups only. Virtually all CRfunctions assume that each unit decrease in the ambi-ent concentration of a pollutant results in a fixed per-centage change in the cases of illness or deathsavoided, independent of the initial pollution level.This assumption may not be valid when ambient con-centration levels are several-fold higher than in citieswhere studies have been conducted, as is the casewhen applying CR functions estimated in industrialcountries for fine particles to cities in developingcountries.

Ideally, cities considering significant policychanges to address air pollution problems shouldconduct an epidemiological study locally. In practice,the complexity and costs of undertaking these studieshave limited their number. Instead, cities typicallytransfer information on health impacts of pollutantson the susceptible demographic groups from existingstudies conducted elsewhere. Box 6 gives an example

A N N E X8


Human Health Effects of the Common Air Pollutants

Pollutant Quantified health effects Unquantified health effects Other possible effects

Ozone Mortality Increased airway responsiveness Immunologic changesMorbidity: to stimuli Chronic respiratory diseases

Respiratory symptoms Extrapulmonary effects (changes inMinor RADs the structure or function of theRespiratory RADs organs)Hospital admissionsAsthma attacksChanges in pulmonary functionChronic sinusitis and hay feverCentroacinar fibrosisInflammation in the lung

Particulate Mortality Changes in pulmonary function Chronic respiratory diseases othermatter/sulfates Morbidity: than chronic bronchitis

Chronic and acute bronchitisHospital admissionsLower respiratory illnessUpper respiratory illnessChest illnessRespiratory symptomsMinor RADsAll RADsDays of work lossModerate or worse asthma status (asthmatics)Inflammation in the lung

CO Morbidity: Behavioral effects Other cardiovascular effectsHospital admissions— Other hospital admissions Developmental effects congestive heart failureDecreased time to onset of angina

NOx

Morbidity: Increased airway responsiveness Decreased pulmonary functionRespiratory illness Inflammation of the lung

Immunological changes

SO2

Morbidity in exercising asthmatics: Respiratory symptoms in non-Changes in pulmonary function asthmaticsRespiratory symptoms Hospital admissions

Lead Mortality Health effects for individuals in ageMorbidity: ranges other than those studied

Hypertension Neurobehavioral functionNonfatal coronary heart disease Other cardiovascular diseasesNonfatal strokes Reproductive effectsIntelligence quotient (IQ) loss Fetal effects from maternal effect on lifetime earnings exposureIQ loss effects on special Delinquent and antisocial education needs behavior in children

Source: U.S. EPA 1997b.

T A B L EA 8. 1

ANNEX 8. ESTIMATING THE HEALTH IMPACTS OF AIR POLLUTION 153

of a CR function transferred in a health impact esti-mation study of Mexico City (Mexico Air QualityManagement Team 2002). Similar functions are avail-able for other health impacts from PM10 as well asother pollutants such as ozone. The appropriatenessof transferring these functions depends on whetherthe confounding factors for the city are similar tothose for the cities included in the transferred epide-miological studies.

The second step in health impact estimation re-quires two pieces of information about the city: (1) thebaseline cases of illness or death; and (2) the change inthe population exposure to the pollutant. Baseline

Estimating a Health Impact ofLowering PM10 Concentrations

Estimating the impact of lowering ambient PM10 concentrations on

avoided hospital admissions for respiratory problems is given below,

taken from a study in Mexico City.

Step 1: Epidemiological study (transferred)� Demographic groups: all

� Concentration-response relationship: 0.139 percent change inhospital admissions for a change in the daily average PM10 con-centration of 1 µg/m³

Step 2: Data from Mexico City� Population at risk: 18,787,934 persons

� Baseline rate of hospital admissions for respiratory problems: 411admissions per 100,000 persons

� Baseline number of hospital admissions: Population at risk ×0.00411 admissions/person = 77,218 admissions

� Current population-weighted annual average PM10 concentration:64 µg/m³

� Population-weighted annual PM10 concentration after policy imple-mentation: 51.2 µg/m³

� Change in PM10 concentration in response to policy implementa-tion: 64 µg/m3 – 51.2 µg/m³ = 12.8 µg/m³

Step 3: Calculation of estimated avoided casesAvoided hospital admissions: 77,218 admissions × 0.00139 change/

µg/m³ × 12.8 µg/m³ = 1,376 admissions

Note: Reduced hospital admissions are only one of the healthbenefits of reducing PM10 concentrations. Other impacts includepremature death and less serious illnesses not requiring hospital-ization.

B O X 6

cases are typically estimated from the total populationand the case incidence rate. The change in the popula-tion exposure is the difference between the current ex-posure level and estimates of population exposurelevels after air pollution reductions are achieved.

Pollutant exposure levels are difficult to estimatebecause of varying personal time-activity patterns. Asa result, health impacts are generally based on thepopulation-weighted average ambient concentrationof the pollutant across the city’s susceptible residents.These ambient levels are estimated from the concen-trations of the pollutants measured at fixed monitor-ing sites located in different parts of the city. It is im-portant to ensure that the monitoring sites arerepresentative of average exposure and are not un-duly influenced by pollution “hot spots” such as in-ner-city transport corridors or industrial zones.

The estimated burden of disease from air pollu-tion, such as 800,000 deaths annually in the world re-ported in a recent publication by the World HealthOrganization (WHO 2002), provides a useful bench-mark for comparing the relative magnitude of differ-ent health risk factors. However, these deaths andother health estimates are not an appropriate basis forcomparing different air pollution reduction strategies.Burden-of-disease estimates are based on reducing airpollution to the theoretically minimum levels (for ex-ample, PM10 concentration of 15 µg/m3). Pollution re-ductions to such low levels have not been achieved inmany U.S. and European cities, and it would be unre-alistic to assume that many heavily polluted develop-ing country cities are in a position to reach these levelsin the near future. Instead, health gain estimatesshould be determined for each pollution reductionstrategy based on the expected population exposurereductions.

The estimated avoided cases of illness or diseaseare calculated in the third step using the informationcollected in the first two steps. Box 6 illustrates howthe estimated avoided cases of hospital admissionsfor respiratory disease are calculated for Mexico Cityfor a 20 percent reduction in population exposures toPM10. The avoided cases provide a concrete measureof health gains understandable to a wide range ofpolicymakers.


Results from Existing StudiesEpidemiological studies can be grouped according tohow exposure is measured—acute exposure studiesand chronic exposure studies—and how health effectsare measured—individual-based panel (or cohort)studies and population-based (or ecological) studies.Most studies in the scientific literature have examinedacute, not chronic, health consequences.

Human health impacts of acute exposure toparticulate air pollutionAcute exposure studies examine the associations be-tween short-term (daily or multiday average) varia-tions in PM concentrations and short-term counts oftotal deaths, cause-specific deaths, or incidence ofspecific illness in an area (typically a city). The popu-larity of these studies stems from their minimal datarequirement compared with other study designs.Problems associated with confounding are reduced inthese studies because population characteristics (suchas smoking and occupational exposures) do notchange much over the study period for the studypopulation. In addition to air pollution, temporal andmeteorological conditions and the age of the indi-vidual are the main factors that are included in thesestudies. While these studies provide health impact es-timates for the city being studied, the CR functionsobtained are not readily transferable to cities with dif-ferent population characteristics.

However, the consistent findings across a wide ar-ray of cities, including those in developing countrieswith diverse population and possibly PM characteris-tics, strongly indicate that the health gains indeed re-sult from PM pollution reductions. Meta-analysis—which pools results from several studies—of acuteexposure studies provides health impact estimatesthat are more transferable than results from indi-vidual studies. These results indicate that every 10µg/m³ increase in the daily or multiday average con-centration of PM10 increases (1) non-trauma deaths by0.8 percent; (2) hospital admissions for respiratoryand cardiovascular diseases by 1.4 and 0.6 percent, re-spectively; (3) emergency room visits by 3.1 percent;(4) restricted activity days by 7.7 percent; and (5)

cough with phlegm in children by 3.3 to 4.5 percent(Cohen and others 2003, Holgate and others 1999).The studies also indicate higher risk for the elderlywith chronic heart and lung disease and infants.

Human health impacts of chronic exposure toparticulate air pollutionChronic exposure studies examine the impact of long-term exposure to particulate air pollution as well asthe cumulative effects of short-term elevated PM lev-els. These studies compare differences in health out-comes across several locations at a selected period intime. Some portion of the long-term impacts indicatedby these studies corresponds to the impact of acute ef-fects revealed in acute exposure studies. The remain-der is caused by latent or chronic effects of cumula-tive exposure.

Ecological studies, which use population-widemeasures of health outcomes, have consistently foundincreased mortality rates in cities with higher PM lev-els. However, the inability to isolate the effects of PMfrom alternative explanatory factors (that is, con-founding factors such as smoking, dietary habits, age,and income) that might vary among populations indifferent cities raises doubts about the reliability ofthese CR functions.

Cohort design studies overcome these questionsby following a sample of individuals, thereby makingit easier to isolate the effects of confounding factors.These studies provide the most compelling evidenceabout mortality effects from chronic exposure to PM.The largest study to date (Pope and others 2002) indi-cates that a change in long-term exposure to PM2.5 of10 µg/m³ leads to a 4, 6, and 8 percent increase in therisk of all-cause mortality, cardiopulmonary mortality,and lung cancer mortality, respectively. The study didnot find consistent relationships between long-termexposure to particles larger than 2.5 microns and pre-mature death.

Estimating Health Effects in DevelopingCountriesOnly a few studies based on measured ambient con-centrations of PM10 or PM2.5 have been carried out in

ANNEX 8. ESTIMATING THE HEALTH IMPACTS OF AIR POLLUTION 155

developing countries. Quantitative estimates of healthgains in the immediate future will have to rely on thetransfer of CR functions. Uncertainties about thesetransfers due to confounding need to be addressedthrough scenario-based sensitivity analysis.

Because health risks from PM affect primarily theelderly with chronic heart and lung diseases and in-fants, transfer of cause- and age-specific CR functionsis preferable. Use of all-cause or all-age mortality isinappropriate when there are systematic differencesin other health risks or the age distribution betweenthe population in the city and those used in the epide-miological studies. For example, cardiovascular andrespiratory diseases have been reported to account fora quarter of non-trauma deaths in Delhi, comparedwith half in the United States (Cropper and others1997). If the cardiopulmonary-specific CR functionfrom the study by Pope and others (2002) were trans-ferred to Delhi, all-cause mortality would increase by1.5 percent when PM2.5 exposure is increased by 10µg/m3, compared with a 4 percent increase if the all-cause CR function from the same study were applied.

Three CR functions transferred to cities world-wide by WHO in one of its publications (Cohen andothers 2003) can be a basis for CR function transfers todeveloping country cities. They include two separateCR functions for cardiopulmonary and lung cancermortality for adults over 30 years of age from chronicexposure and a CR function for all-cause mortality inchildren from acute exposure. No morbidity CR func-tions were transferred because definitions of healthoutcomes differ across countries. The economic lossesfrom the morbidity effects of PM pollution are suffi-ciently significant that excluding them would seri-ously underestimate the cost of air pollution. CR func-tions for morbidity can be transferred, provided thatthe differences in the confounding factors and defini-tions of health outcomes between the developingcountry cities and those in the epidemiological stud-ies are properly accounted for.

Uncertainty from three additional sources shouldbe addressed through sensitivity analysis: lack of dataon fine PM concentrations, lack of baseline healthdata and cases, and extrapolation of CR functions out-

side of the pollutant concentration ranges observed inthe epidemiological studies. A number of developingcountry cities have historically monitored total sus-pended particles (TSP). Recently, quite a few citieshave begun to monitor PM10 regularly, and some aremonitoring PM2.5. Because the size distribution of PMvaries significantly depending on the sources of pol-lution and atmospheric conditions, estimating theconcentration of fine particles in the absence of locallymeasured data is not straightforward. A World Bankstudy (Pandey and others 2003) found that after con-trolling for the fuel mix and local climatic factors,PM10 accounts for a smaller share of TSP as per capitaincome falls.

Ambient PM concentrations are significantlyhigher in many developing country cities than thosefound in epidemiological studies in North Americaand Europe, requiring extrapolation of CR functionsabove the maximum PM concentrations found in theoriginal epidemiological studies. If CR functions werelinearly extrapolated, then at high particulate levelsfound in some developing country cities, a significantproportion of health outcomes would be estimated tobe from exposure to PM rather than from other com-peting factors such as smoking and high blood pres-sure. Since little collaborating evidence has beenfound to support such a conclusion, it would seemmore reasonable to assume that, at these higher levels,the additional health impact per unit µg/m3 increasein exposure would be smaller. Different assumptionsabout extrapolation can be used to estimate high, cen-tral, and low estimates of health effects, as shown in arecent WHO publication (Cohen and others 2003).

Conclusions� The health impacts of air pollution depend on

the sensitivity and the exposure level of the sus-ceptible population to the pollutant. The largesthealth impacts in most developing country citiesresult from exposure to fine particulate pollu-tion. Elderly persons with cardiovascular andlung disease and infants are at greatest risk.

� In performing health impact analyses for mostdeveloping country cities, reliance has to be


placed on CR transfer in the immediate future.For mortality, this should be limited to CR func-tions for cause- and age-specific mortality devel-oped from PM10 and PM2.5 measurements. CRfunctions for morbidity can be transferred, pro-vided that definitions of health outcomes arecomparable and there are no large differences inconfounding factors.

� Uncertainties in transferring CR functionsshould be fully addressed by examining the sen-sitivity of results to alternative assumptions. Themost significant uncertainties are baseline healthdata, estimations of PM2.5 and PM10 needed inthe CR function if no local data exist, and ex-trapolation of CR function outside of the PMconcentration range in the original studies.

157

A N N E X9

Valuing Health Effects

Reductions in ambient levels of common air pollut-ants have been associated with reductions in prema-ture mortality from heart and lung disease as well asreductions in chronic bronchitis, asthma attacks, andother forms of respiratory illness. This annex presentsthe methods used to perform economic valuation ofchanges in illness and premature mortality and dis-cusses the appropriateness of transferring health ben-efit estimates from studies in other regions to devel-oping countries. This is followed by illustrations ofhow the monetary value of health benefits associatedwith improvements in air pollution can be useful topolicymakers (World Bank 2003b).

Valuing Reductions in Illness

What is being valued

Improving air quality should reduce the number ofepisodes of acute illness (such as asthma attacks) aswell as the number of cases of chronic respiratory ill-ness that occur each year. To economists, the value ofavoiding an illness episode, such as an asthma attack,consists of four components: (1) the value of the worktime lost due to the attack (by the asthmatic or an un-paid caregiver or both); (2) the medical costs of treat-ing the attack; (3) the amount an asthmatic (or, in thecase of a child, the child’s guardians) would pay toavoid the pain and suffering associated with the at-tack; and (4) the value of the leisure time lost due tothe attack (by the asthmatic or a caregiver).

If the asthmatic were to bear all costs of the attack(including lost work time and medical costs), his orher stated willingness to pay should reflect all fourcomponents of value. If, in contrast, the asthmatic hadhealth insurance and paid sick leave, he or shewould not bear all medical costs and productivitylosses. These are, however, legitimate economic

costs that must be included in the value of an ill-ness episode.

Calculating the value of avoided illnessHow are the four components of the value of avoid-ing illness measured? Medical costs and productivitylosses are often estimated by asking about the type oftreatment sought during an illness episode and byasking how long the episode lasted and for howmany days the patient (or a family caregiver or both)were unable to perform their usual duties. Lost worktime is then valued at the wage rate, and medicalcosts are imputed on the basis of the full social costsof providing the care, not just the costs to the patient.Economists usually estimate the value of pain andsuffering avoided and the value of leisure time gainedby direct questioning: that is, people are asked whatthey would pay to avoid the discomfort and inconve-nience of an illness of a specific type and duration.This approach is referred to as the contingent valuation

method (CVM) or the stated preference method.

When estimates of the value of pain and sufferingand lost leisure time are unavailable, medical costsand productivity losses are often used to provide alower bound to the value of avoiding illness. This isreferred to as the cost-of-illness (COI) approach tovaluing morbidity. Medical costs are referred to as thedirect costs of illness, and productivity losses as the in-

direct costs of illness. In the case of a serious but infre-quent illness, such as a stroke, reducing air pollutionreduces the risk of a person having a stroke. Thuswhat should be estimated is what a person would payto reduce his or her risk of having a stroke. In practice,the COI approach is often used to value serious ill-nesses, such as a heart attack or stroke, since empiri-cal estimates of what people are willing to pay toavoid the pain and discomfort of these conditionstend to be lacking.


Valuing Reductions in PrematureMortality

What is being valuedStudies of the effects of air pollution on prematuremortality predict how many fewer people are likely todie if air pollution is reduced. For example, a 10 per-cent reduction in PM10 in Delhi, India, might result in1,000 fewer deaths each year. We refer to the 1,000fewer deaths as the number of statistical lives saved byimproving air quality. This means that the risk of dy-ing is reduced by a small amount for all people livingin Delhi and that these risk reductions add up to 1,000fewer deaths. To illustrate, if reducing air pollution inDelhi results in 1,000 fewer deaths in a population of10 million, this is equivalent, on average, to reducingrisk of death annually by 1 in 10,000 (0.0001) for eachperson in the population (calculated from dividing1,000 deaths by 10 million people, or 0.0001).1

Since reducing air pollution reduces risk of deathby a small amount for each person in an exposedpopulation, economists wish to estimate what eachperson in the population would pay for this small riskreduction. If this willingness to pay (WTP) wereadded across all 10 million residents of Delhi, itwould represent the value of saving 1,000 statisticallives. Dividing the total willingness to pay by thenumber of statistical lives saved yields the averagevalue of a statistical life (VSL). People’s WTP for smallrisk reductions are usually stated in terms of theVSL—the sum of WTPs for risk reductions that saveone statistical life.2

Calculating the value of a reduction inrisk of deathEconomists realize that people trade money for safetyevery day. People are willing to work in riskier jobs ifcompensated for them, and people are willing to payfor safer vehicles or for helmets to protect themselves

when riding two-wheelers. WTP for a reduction inrisk of dying is usually estimated from studies oncompensating wage differentials in the labor market,or expenditures to reduce risk of death. These studiesare usually referred to as revealed preference studies be-cause they are based on actual behavior. A secondsource of estimates are stated preference studies inwhich people are asked directly what they would payfor a reduction in their risk of dying (also called CVMand referred to above in the context of valuing mor-bidity).

Studies of compensating wage differentials or ex-penditures on safety must determine what portion ofthe wage or what portion of the vehicle price repre-sents payment for safety. This payment is then associ-ated with the size of the risk differential to infer whatpeople are willing to pay for it. For example, compen-sating wage studies empirically explain variations inthe wage received by workers as a function of workercharacteristics (age, education, skills) and job charac-teristics, including risk of fatal and nonfatal injury, inorder to determine what portion of wage representscompensation for risk of death. In theory, the impactof small changes in the risk of dying on wages shouldequal the amount a worker would have to be com-pensated to accept this risk.

Compensating wage differential studies in theUnited States indicate that the VSL is approximately$5 million (1990 US$) (U.S. EPA 1999f). These studiesmay overstate the VSL for reductions in air pollutionbecause people prematurely dying from air pollutionin North America are much older than the workers inthese studies, whose average age is about 40. Con-versely, the VSL for environmental risks may behigher because these risks are involuntary.

Unlike compensating wage differential studies,contingent valuation studies directly ask persons at riskwhat they are willing to pay for changes in life expect-ancy, and can be tailored to the age at which risk re-ductions occur and to the nature of the risks valued.They generally yield lower estimates of WTP thanwage differential studies do. These studies often havedifficulty eliciting consistent values for small prob-ability changes that are difficult for respondents toperceive and value.

1For simplicity, this example assumes that all people in Delhibenefit equally from the air pollution reduction. In reality, peoplewith heart and lung disease are likely to benefit more than others.

2The goal of calculating the VSL is to estimate what peoplethemselves would pay for risk reductions. The VSL is not in-tended to estimate the intrinsic value of human life.

ANNEX 9. VALUING HEALTH EFFECTS 159

When WTP estimates are not available, the human

capital (“human capital” refers to knowledge andskills found in the labor force) approach can be usedto obtain a lower bound to WTP. This approach val-ues loss of life based on the forgone earnings associ-ated with premature mortality. The notion is thatpeople should be willing to pay at least as much asthe value of the income they would lose by dying pre-maturely. This is not the theoretically correct ap-proach to valuing a program that reduces the risk ofdying, but does provide a useful lower bound to WTP(Freeman 2003). Labor market studies in the UnitedStates indicate the VSL is several times the value offorgone earnings. However, caution is urged in usingthese values for policy formulation: if the same VSL isused for reducing a number of different risks, addingup estimates of WTP could result in an unrealisticallylarge monetary sum that is out of proportion to the to-tal household income for the majority of the popula-tion.

Valuing Health Benefits in DevelopingCountriesFew studies have been published using data for de-veloping countries that estimate WTP to reduce mor-tality or morbidity. This implies that monetization ofhealth benefits must, in the immediate future, rely ontransferring WTP estimates from one country to an-other or must calculate a lower bound to benefitsbased on forgone earnings (for mortality benefits) orthe cost of illness (for morbidity benefits).

The standard approach to benefits transfer as-sumes that preferences are the same in the two coun-tries, including attitudes toward risk when estimatesof the VSL are transferred. WTP is assumed to differonly as a result of differences in income between thetwo countries. If this is true, U.S. WTP can be trans-ferred to a specific developing country after account-ing for income differences as shown in equation (1):

WTPDC = WTPUS [IncomeDC/IncomeUS]e(1)

where IncomeDC is the income in the developingcountry measured in U.S. dollars and e represents theincome elasticity of WTP: the percentage change inWTP corresponding to a 1 percent change in income.

There is considerable uncertainty regarding the in-come elasticity of WTP, even within a country. A con-servative approach to benefits transfer is to use an in-come elasticity of 1.0, including smaller and largervalues for sensitivity analysis. For example, the trans-fer of a U.S. VSL of US$1 million to India using 1998purchasing power parity3 income and an elasticity of1.0 yields a VSL for India of US$69,000. Using thenominal exchange rate to convert the income in Indiarather than purchasing power parity would give alower estimate, and is typically not done on method-ological grounds. Since the assumptions underlyingbenefits transfer may not be valid, it is always desir-able to provide lower-bound estimates of the value ofhealth benefits based on the COI approach for mor-bidity and the human capital approach for mortalityand to compare these with higher values based on theWTP and VSL approaches.

The Policy Relevance of Health-BenefitsAnalysis—Example from Mexico CityTo illustrate the usefulness of computing the mon-etary value of health benefits, the results of a study inMexico City are given in table A9.1. The study quanti-fied the effect of 10 and 20 percent reductions in an-nual average population-weighted ozone and PM10

concentrations in metropolitan Mexico City in theyear 2010. The impact of each pollutant reduction wasfirst expressed in terms of cases of illness and prema-ture death avoided; then dollar values were assignedto health benefits.

Three approaches were used to value reductionsin illness and premature mortality. The most conser-vative, giving the “low estimate,” was to value mor-tality using forgone earnings and morbidity usingproductivity losses plus medical costs, that is, COI.This should be viewed as a lower bound to the valueof health benefits. A less conservative approach, giv-

3Purchasing power parity is a way to compare the costs ofgoods and services between countries. It is different from the offi-cial exchange rate and considers a rate of exchange that will giveeach currency the same purchasing power in its economy. Interms of purchasing power parity, local currencies in many devel-oping countries are worth much more than the official exchangerate.


ing the “central case estimate,” was to add estimatesof WTP to avoid the pain and suffering associatedwith illness to the COI used in computing the low es-timate of benefits, but to use forgone earnings tovalue reduced mortality. The “high estimate” used thesame method of valuing avoided morbidity as thecentral case estimate but uses WTP (that is, the VSL)in place of forgone earnings to value avoided mortal-ity.

WTP estimates were transferred from studies con-ducted in the United States and Europe, using an in-come elasticity of WTP of unity and purchasingpower parity incomes. The resulting VSL for MexicoCity was approximately $300,000 in 1999 US$.

The values of reducing PM10 and ozone by 10 per-cent and 20 percent appear in table A9.1. Two featuresof the results warrant discussion. The first is that thedollar values of benefits associated with the 20 per-cent reduction scenario are exactly twice the values ofthe benefits of the 10 percent reduction scenario. Ingeneral, each additional 1 µg/m3 reduction in annualaverage population-weighted PM10 will have approxi-mately the same value, because the impact of a 1 µg/m3 change is assumed to be independent of baselineconcentrations. In this case a reduction of 10 percent isequivalent to lowering PM10 by 6.4 µg/m³. This im-plies that, using the study’s central case estimate, a 1µg/m3 reduction in PM10 is worth $100 million (1999

US$) annually (US$644 million for COI+WTP dividedby 6.4 µg/m³).

The second point worth noting is that the value ofthe health benefits associated with a 10 percent reduc-tion in ozone is much smaller than the value of thebenefits of a 10 percent reduction in PM10 regardlessof the approach used to monetize benefits. This is pri-marily because there are no studies, as yet, relating re-ductions in long-term exposure to ozone to prematuremortality. This does not imply, however, that pro-grams to reduce the precursors of ozone (NOx andVOCs) yield few health benefits. In addition, NOx andSOx can convert to secondary particulate matter in theatmosphere. Programs to reduce oxides of nitrogenand sulfur are, therefore, likely to result in benefitsfrom reduced particulate matter.

The Use of Benefit Estimates inCost-Benefit AnalysesThe estimates of health benefits, such as those com-puted in the Mexico City study in table A9.1, could beused as inputs to a cost-benefit analysis of air pollu-tion control strategies. To analyze the benefit of an airpollution control strategy, one must first translate thecontrol measures—for example, a program to convertdiesel buses to CNG—into changes in emissions ofthe common air pollutants, and then use air qualitymodels to predict the change in ambient pollutionconcentrations associated with the control strategy.Once the changes in ambient concentrations associ-ated with the control strategy have been estimated,they can be quantified and valued using the unit val-ues derived in the health-benefits analysis.

The final step in a cost-benefit analysis is to sub-tract the costs of the program (such as the cost of re-placing diesel buses with CNG buses) from its ben-efits to determine the net social benefits of theprogram. Economists typically argue that controlstrategies should be ranked according to their net so-cial benefits; this assumes that what matters are thetotal benefits to society versus the total costs to societyof a program, even if the people who pay for the pro-gram are not the same people as those who benefitfrom it. The distribution of benefits and costs is, how-

Annual Health Benefits due to Ozone and PM10

Reductions in Mexico City (million 1999 US$)

Air pollutionMethodology for calculation reduction

Morbidity Mortality 10% 20%

Benefits from ozone reductionCOI Human capital 18 35COI + WTP Human capital 75 151COI + WTP VSL 116 232

Benefits from PM10 reductionCOI Human capital 96 191COI + WTP Human capital 644 1,289COI + WTP VSL 1,451 2,903

Source: Mexico Air Quality Management Team 2002.

T A B L EA 9. 1

ANNEX 9. VALUING HEALTH EFFECTS 161

ever, important information that should also be pre-sented to policymakers, in addition to total benefitsand costs.

Conclusions� The economic benefits of reducing illness and

premature mortality associated with air pollu-tion are well defined, and empirical estimates ofthese benefits (for example, of the VSL) exist forindustrial countries.

� In performing health-benefits analyses for mostdeveloping countries, reliance will have to beplaced on benefits transfer in the immediate fu-ture. In addition, it should be possible to calcu-

late a lower bound to benefits using the cost ofillness and human capital approaches. Policy in-terventions that can be justified on the basis oflower-bound estimates of benefits are likely tobe robust and merit serious consideration.

� Calculating the monetary value of health ben-efits associated with small (for example, 10 per-cent) changes in the common air pollutants isuseful for two reasons: (1) it provides estimatesof the value of a one-unit reduction in each pol-lutant that can serve as input into a cost-benefitanalysis of air pollution reduction strategies; and(2) it can indicate the relative benefits of control-ling one pollutant versus another.

163

References

ACEA (Association des Constructeurs Européens d’Automobiles)and Europia. 1995. European Programme on Emissions, Fuelsand Engine Technologies: EPEFE Report.

ACEA (Association des Constructeurs Européens d’Automobiles),Alliance of Automobile Manufacturers, EMA (EngineManufacturers Association), and JAMA (Japan AutomobileManufacturers Association). 2002. “World-Wide Fuel Char-ter.” December. Available online [April 26, 2004] at:www.autoalliance.org/archives/000090.html.

Airborne Particles Expert Group. 1999. “Source Apportionment ofAirborne Particulate Matter in the United Kingdom.” Janu-ary. Available online [April 26, 2004] at:www.defra.gov.uk/environment/airquality/airbornepm/.

Alternative Fuels Contact Group. 2003. “Market Development ofAlternative Fuels.” December. Available online [April 26,2004] at: http://europa.eu.int/comm/energy_transport/envir/2003_report_en.pdf.

Asian Development Bank. 2003. “Cleaner Fuels: Policy Guidelinesfor Reducing Vehicle Emissions in Asia,” Manila. Availableonline [April 26, 2004] at: www.adb.org/documents/guidelines/Vehicle_Emissions/cleaner_fuels.pdf.

ASTM (American Society for Testing and Materials). 1997. AnnualBook of ASTM Standards. West Conshohocken, Pennsylvania.

Auto/Oil Air Quality Improvement Research Program. 1997. Pro-gram Final Report.

Automotive Environment Analyst. 2000. “Seattle AFVs Prove TooMuch Hassle.” March, 62: 3.

———. 2001. “Texaco Researches Cheaper Drilling,” September,80: 7.

———. 2003a. “…But ACEA Boss Calls CO2 Goals Impractical.”December 2003–January 2004, 104: 9.

———. 2003b. “China to Enact Fuel Standards.” December 2003–January 2004, 104: 17.

———. 2004. “Mixed Responses to Bush CAFE Proposals.” Febru-ary, 105: 23.

Bacon, Robert W. 2001. “Oil Product Taxes.” Viewpoint 240. Sep-tember. Washington, D.C.: World Bank. Available online[April 26, 2004] at: www.worldbank.org/ogmc/files/240Bacon-831.pdf.

Bayliss, D. 1989. Background report for the European Conferenceof Ministers of Transport-OECD Joint Ministerial Sessionon Transport and the Environment, Paris, November.

Beer, Tom, Tim Grant, Geoff Morgan, Jack Lapszewicz, PeterAnyon, Jim Edwards, Peter Nelson, Harry Watson, andDavid Williams. 2001. “Comparison of Transport Fuels.”Final Report (EV45A/2/F3C) submitted to the AustralianGreenhouse Office on the Stage 2 study of Life-cycle Emis-sions Analysis of Alternative Fuels for Heavy Vehicles.Available online [April 26, 2003] at:www.greenhouse.gov.au/transport/comparison/.

Betts, W.E., S.Å Fløysand, and F. Kvinge. 1992. “The Influence ofDiesel Fuel Properties on Particulate Emissions in Euro-pean Cars.” SAE Technical Papers Series 922190.Warrendale, Pennsylvania: Society of Automotive Engineers.

Bharat Petroleum. Undated. “Pure for Sure.” Available online[April 26, 2004] at: www.bharatpetroleum.com/retail/retail_pure.asp.

Blumberg, Katherine O., Michael P. Walsh, and Charlotte Pera.“Low-sulfur Gasoline & Diesel: The Key to Lower VehicleEmissions.” Available online [April 26, 2004] at:www.cleantransportcouncil.org/docs/Sulfur_Report.pdf.

bp. Statistical Review of World Energy 2003. Available online [April26, 2004] at: www.bp.com/subsection.do?categoryId=95&contentId=2006480.

Bunting, Alan. 2003. “Oil and Fuel Additives Come Under ‘Emis-sion Effects’ Scrutiny.” Automotive Environment AnalystMay, 98: 23–25.

Cambridge Economic Policy Associates. 2002. “Sri Lanka FuelStudy.” A report prepared for the Government of SriLanka, November.

CARB (California Air Resources Board). 2000a. “Fact Sheet: Cali-fornia Diesel Fuel.” October 6. Available online [April 26,2004] at: www.arb.ca.gov/fuels/diesel/dieselfs.pdf.

———. 2000b. “Final Resolution Order.” Available online [April26, 2004] at: www.arb.ca.gov/regact/motorcyc/finreg.pdf.

———. Undated. “Diesel Fuel Program.” Available online at:www.arb.ca.gov/fuels/diesel/diesel.htm.

Chandra, Ajay, Ronald L. Gist, Ken W. Otto, and S. Craig Whitley.2001. “LPG Trade Patterns Continue Historic Change in2000.” Oil and Gas Journal June 25: 75–79.

Cohen, A., R. Anderson, B. Ostro, K.D. Pandey, M. Krzyzanowski,N. Kunzli, K. Gutschmidt, A. Pope, I. Romieu, J. Samet,and K. Smith. 2003. “Mortality Impacts of Particulate AirPollution in the Urban Environment.” In M. Ezzati, A.D.Lopez, A.D. Rodgers, and C.J.L. Murray, eds., ComparativeQuantification of Health Risks: Global and Regional Burden ofDisease due to Selected Major Risk Factors. Geneva: WorldHealth Organization.

Concawe. 1997. “Motor Vehicle Emission Regulations and FuelSpecifications. Part 2. Detailed Information and HistoricReview (1970–1996).” Report No. 6/97, March, Brussels.

Cracknell, J.A. 2000. “Experience in Urban Traffic Managementand Demand Management in Developing Countries.”Background Paper to the World Bank Urban TransportStrategy Review. Available online [April 26, 2004] at:www.worldbank.org/transport/utsr/background_papers/uk_traffic_manag_cracknell.pdf.

Cropper, M., N. Simon, A. Alberini, and P.K. Sharma. 1997. “TheHealth Effects of Air Pollution in Delhi, India.” World BankPolicy Research Working Paper 1860. Washington, D.C.:World Bank. Available online [April 26, 2004] at: http://econ.worldbank.org/docs/534.pdf.


Dieselnet. Undated. “Emission Standards.” Available online[April 26, 2004] at: www.dieselnet.com/standards.html.

Durbin, T.D., M.R. Smith, J.M. Norbeck, and T.J. Truex. 1999.“Population Density, Particulate Emission Characterizationand Impact on the Particulate Inventory of Smoking Ve-hicles in the South Coast Air Quality Management Dis-trict.” Journal of Air and Waste Management Association49:28–38.

ECMT (European Conference of Ministers of Transport). 1999.“Improving the Environmental Performance of Vehicles:Fleet Renewal and Scrappage Schemes.” Report by the Eu-ropean Conference of Ministers of Transport Committee ofDeputies to the Council of Ministers. CEMT/CS(99)13.Available online [April 26, 2004] at: www.olis.oecd.org/olis/1999doc.nsf/0/88633be106cfd8e2c12567450043809b?OpenDocument.

The Economist. 2004. “Road Traffic Deaths.” April 17–23: 102.

Englebrecht, J.P., and L. Swanepoel. 1998. “Aerosol Monitoringand Source Apportionment in Qalabotjha, Free State, SouthAfrica.” Report No. MC5 prepared for Mintek.

Environment Directorate, Environment Policy Committee. 2002.“Synthesis of OECD Work on Environment and Transportand Survey of Related OECD, IEA and ECMT Activities.”April 11, ENV/EPOC/WPNEP/T(2002)7/FINAL.

Environment Protection and Heritage Council. Diesel Vehicle Emis-sions Preparatory Projects. Available online [April 26, 2004]at: www.ephc.gov.au/nepms/diesel/diesel_prepare.html.

ESMAP (Energy Sector Management Assistance Programme).2001a. Chapter 4, “Mitigating Vehicular Emissions.” InMexico Energy Environment Review. ESMAP Report 241/01,May. Washington, D.C. Available online [April 26, 2004] at:www.worldbank.org/html/fpd/esmap/publication/mexicoenergyreview.html.

———. 2001b. Pakistan Clean Fuels. ESMAP Report 246/01, Octo-ber. Washington, D.C. Available online [April 26, 2004] at:www.worldbank.org/html/fpd/esmap/pdfs/pakistan_cleanfuels_intro.pdf.

———. 2002a. Chapter 2, “Measuring Particulate Emissions.” InBangladesh: Reducing Emissions from Baby-Taxis in Dhaka.ESMAP Report 253/02, January. Washington, D.C. Avail-able online [April 26, 2004] at: www.worldbank.org/html/fpd/esmap/pdfs/253-02.pdf.

———. 2002b. Latin America and the Caribbean Refinery Sector De-velopment Project and Latin America and the Caribbean Refin-ery Sector Development Project—Refinery Sector Report Annex.ESMAP Technical Paper 026-02, August. Washington, D.C.Available online [April 26, 2004] at: www-wds.worldbank.org/servlet/WDS_IBank_Servlet?pcont=details&eid=000094946_02122004122024 and http://wbln0018.worldbank.org/esmap/site.nsf/files/226-02+LAC+-+Refinery+Sector+Development+Project+-+Annex+-++Vol+III.pdf/$FILE/226-02+LAC+-+Refinery+Sector+Development+Project+-+Annex+-++Vol+III.pdf.

———. 2004. Chapter 2, “Controlling Emissions from In-Use Ve-hicles.” In Toward Cleaner Urban Air in South Asia: TacklingTransport Pollution, Understanding Sources. ESMAP Report281/04, March. Washington, D.C. Available online [April26, 2004] at: www.worldbank.org/sarurbanair.

Eudy, Leslie. 2002. “Natural Gas in Transit Fleets: A Review of theTransit Experience.” February. Golden, Colorado: NationalRenewable Energy Laboratory. Available online [April 26,2004] at: www.ott.doe.gov/otu/field_ops/pdfs/nat_gas_transit_fleets.pdf.

European Commission. 2000a. “The Auto-Oil II Programme: AReport from the Services of the European Commission.”Report by the Directorates General for: Economic and Fi-nancial Affairs, Enterprise, Transport and Energy, Environ-ment, Research and Taxation and Customs Union, October.Available online [April 26, 2004] at: www.europa.eu.int/comm/environment/autooil/auto-oil_en.pdf.

———. 2000b. “Proposal for a Directive of the European Parlia-ment and of the Council Amending Directive 97/24/EC onCertain Components and Characteristics of Two- or Three-Wheel Motor Vehicles.” Available online [April 26, 2004] at:http://europa.eu.int/eur-lex/en/com/pdf/2000/en_500PC0314.pdf.

———. 2001. “Communication from the Commission to the Euro-pean Parliament, the Council, the Economic and SocialCommittee and the Committee of the Regions on Alterna-tive Fuels for Road Transportation and on a Set of Mea-sures to Promote the Use of Biofuels.” COM(2001)547, 7November, Brussels. Available online [April 26, 2004] at:http://europa.eu.int/comm/energy/library/comm2001-547-en.pdf.

European Communities. 1998. “Directive 98/69/EC of the Euro-pean Parliament and of the Council of 13 October 1998 Re-lating to Measures to be Taken Against Air Pollution byEmissions from Motor Vehicles and Amending Council Di-rective 70/220/EEC.” Official Journal of the European Com-munities L350 EN. Available online [April 26, 2004] at:http://europa.eu.int/eur-lex/pri/en/oj/dat/1998/l_350/l_35019981228en00010056.pdf.

European Parliament. 2000. “Common Position Adopted on 10October 2000 by the Council with a View to the Adoptionof a Directive of the European Parliament and of the Coun-cil Amending Council Directive 70/220/EEC ConcerningMeasures to Be Taken Against Air Pollution by Emissionsfrom Motor Vehicles.” Session Document C5-0522/2000,Brussels, October 22. Available online [April 26, 2004] at:www.europarl.eu.int/commonpositions/2000/pdf/c5-0522-00_en.pdf.

EU (European Union). 2003. “Directive 2003/17/EC of the Euro-pean Parliament and of the Council of 3 March 2003Amending Directive 98/70/EC Relating to the Quality ofPetrol and Diesel Fuels.” Official Journal of the EuropeanUnion L76 EN, March 22. Available online [April 26, 2004]at http://europa.eu.int/eur-lex/pri/en/oj/dat/2003/l_076/l_07620030322en00100019.pdf.

EWG (Environmental Working Group). 2003. “EWG Farm Sub-sidy Database 2.0. Findings.” September 8. Available online[April 26, 2004] at: www.ewg.org/farm/findings.php.

Frailey, Mike, Paul Norton, Nigel N. Clark, and Donald W. Lyons.2000. “An Evaluation of Natural Gas versus Diesel in Me-dium-Duty Buses.” SAE Technical Paper Series 2000-01-2822. Warrendale, Pennsylvania: SAE.

Francchia, Juan Carlos. 2000. “An Overview of the ArgentineNGV Experience.” Presentation made at the World BankWorkshop on Compressed Natural Gas, Washington, D.C.,March 2-3. Available online [April 26, 2004] at: www.iangv.org/html/sources/sources/reports/index.php#wbs.

Freeman, A.M. 2003. The Measurement of Environmental and Re-source Values: Theory and Methods. Washington, D.C.: Re-sources for the Future Press.

GEF (Global Environment Facility). 2002. “Proceedings of theBrainstorming Session on Non-Technology Options forStimulating Modal Shifts in City Transport Systems.”Nairobi, Kenya, March. Washington, D.C.: STAP/GEF.

REFERENCES 165

General Motors, Argonne National Laboratory, BP Amoco,ExxonMobil, and Shell. 2001. “Well-to-Wheel Energy Useand Greenhouse Gas Emissions of Advanced Fuel/VehicleSystems—North American Analysis: Executive SummaryReport.” April. Available online [April 26, 2004] at:www.transportation.anl.gov/ttrdc/pdfs/TA/163.pdf.

Global Refining & Fuels Report. 2003. “‘No Diesel Without Filter’Campaign Seen Winning Battles with Automakers.” May14.

Grønskei, Knut Erik, Frederick Gram, Leif Otto Hagen, SteinarLarssen, Huib Jansen, Xander Olsthoorn, MoestikahadiSoedomo, and Umar F. Achmandi. 1996a. URBAIR UrbanAir Quality Management Strategy in Asia: Jakarta Report.Jitendra Shah and Tanvi Nagpal, eds., October. Washing-ton, D.C.: World Bank. Available online [April 26, 2004] at:www-wds.worldbank.org/servlet/WDSServlet?pcont=details&eid=000009265_3980313101856.

Grønskei, Knut Erik, Frederick Gram, Leif Otto Hagen, SteinarLarssen, Huib Jansen, Xander Olsthoorn, Anil S. Giri,Madan L. Shrestha, and Bimala Shrestha. 1996b. URBAIRUrban Air Quality Management Strategy in Asia: KathmanduValley Report. Jitendra Shah and Tanvi Nagpal, eds., Octo-ber. Washington, D.C.: World Bank. Available online [April26, 2004] at: www-wds.worldbank.org/servlet/WDSServlet?pcont=details&eid=000009265_3980313102044.

GTZ (Deutsche Gesselschaft für Technische Zusammenarbeit).2002. “Module 4a: Cleaner Fuels and Vehicle Technolo-gies.” In Sustainable Transport: A Sourcebook for Policy-makersin Developing Cities. Eschborn.

Gwilliam, K.M., and Zmarak Shalizi. 1999. “Road Funds, UserCharges, and Taxes.” World Bank Research Observer 14 (2):159–85.

Gwilliam, Ken, Robert Bacon, Masami Kojima, and KseniyaLvovsky. 2001. Pollution Management in Focus DiscussionNote Number 11. Transport Fuel Taxes and Urban Air Qual-ity. Washington, D.C.: World Bank. Available online [April26, 2004] at: http://lnweb18.worldbank.org/essd/envext.nsf/51ByDocName/DocumentsandPublicationsIn-FocusNotesDiscussionNote11TransportFuelTaxesandUrbanAirQualityDecember2001.

Ha, Kong. 2002. “Hong Kong Diesel Emissions Control Program.”Paper presented at Better Air Quality 2002, Regional Work-shop on Better Air Quality in Asia and Pacific Rim Cities,Hong Kong, China, December 16–18. Available online[April 26, 2004] at: www.cse.polyu.edu.hk/~activi/BAQ2002/Vehicle/36.doc.

Hahn, R. 1995. “An Economic Analysis of Scrappage.” Rand Jour-nal of Economics 26(2): 222.

Halcrow Fox. 2000. “Review of Urban Public Transport Competi-tion.” Background paper to the World Bank Urban Trans-port Strategy Review, May. Available online [April 26,2004] at: www.worldbank.org/transport/utsr/background_papers/uk_competition_bayliss.pdf

Halcrow Fox and Traffic and Transport Consultants. 2000. “UrbanMass Transit in Developing Countries.” Background Paperto the World Bank Urban Transport Strategy Review. Avail-able online [April 26, 2004] at: www.worldbank.org/trans-port/utsr/background_papers/uk_mass_transit_halcrow.pdf.

Halsnæs, Kirsten, Anil Markandya, Jayant Sathaye, Richard Boyd,Alistair Hunt, and Tim Taylor. Undated. “Transport andthe Global Environment: Accounting for GHG Reductionsin Policy Analysis.” Available online [April 26, 2004] at:www.uneptie.org/energy/act/tp/glob_ov/tpglobov.htm.

Hansen, James, and Larissa Nazarenko. 2004. “Soot Climate Forc-ing Via Snow and Ice Albedos.” Proceedings of the NationalAcademy of Sciences 101(2): 423–428.

Hare, Charles T., and James N. Carroll. 1993. “Reactivity of Ex-haust Emissions from a Small Two-Stroke Engine and aSmall Four-Stroke Engine Operating on Gasoline andLPG.” SAE Technical Paper Series 931540. Warrendale,Pennsylvania: Society of Automotive Engineers.

Harris, Garth. 2000. “Compressed Natural Gas in New Zealand.”Presentation made at the World Bank Workshop on Com-pressed Natural Gas, March 2-3, Washington, D.C. Avail-able online [April 26, 2004] at: www.iangv.org/html/sources/sources/reports/WBPaper2.DOC.

———. 2004. Personal communication.

Hidalgo Guerrero, Dario. 2001. “TransMilenio: The Mass Trans-port System of Bogotá.” Paper prepared for the LatinAmerican Urban Public Transport Congress (KLATPU),Havana, Cuba.

———. 2003. “Bus Rapid Transit: Bogotá’s TransMilenio and Les-sons in Asia.” Presentation made at Better Air Quality2003, December 17–19, Manila, Philippines. Availableonline [April 26, 2004] at: www.cleanairnet.org/baq2003/1496/article-57876.html.

Holgate, S., J. Samet, H. Koren, and R. Maynard, eds. 1999. AirPollution and Health. San Diego, California: Academic Press.

IANGV (International Association for Natural Gas Vehicles). Un-dated. “Latest International NGV Statistics.” Availableonline [April 26, 2004] at: www.iangv.org/jaytech/.

ICCT (International Council on Clean Transportation). “Statementon Fuel Sulfur.” Available online [April 26, 2004] at:www.cleantransportcouncil.org/docs/Sulfur_Statement_English.pdf.

IEA (International Energy Agency). 2002. Bus Systems for the Fu-ture: Achieving Sustainable Transport Worldwide. Paris.

Iyer, N.V. 2003. Personal communication.

———. 2004. Personal communication.

JJ&B Consultants. 2001. “Implementation of Rapid Transit.” Back-ground Paper to the World Bank Urban Transport StrategyReview. Available online [April 26, 2004] at:www.worldbank.org/transport/utsr/background_papers/rapid_transit_bb&j.pdf.

Jonk, Gerie. 2002. “European Environment Bureau (EEB) Back-ground Paper 18-03-2002 On the Use of Biofuels for Trans-port.” Available online [April 26, 2004] at: www.eeb.org/publication/EEB-Biofuels-background-18-03-02.pdf.

Kenworthy, J.R., F.B. Laube, and P. Newman. 1999. An Interna-tional Sourcebook of Automobile Dependence in Cities, 1960–1990. Boulder: University Press of Colorado.

Kojima, M., C. Brandon, and J. Shah. 2000. “Improving Urban AirQuality in South Asia by Reducing Emissions from Two-Stroke Engine Vehicles.” December. Washington, D.C.:World Bank. Available online [April 26, 2004] at: http://lnweb18.worldbank.org/SAR/sa.nsf/Attachments/ImprovUrbanAirQual/$File/Improving+Urban+Air+Quality.pdf.

Kojima, Masami, and Robert Bacon. 2001a. “Abuses in Fuel Mar-kets.” Viewpoint 237, September. Washington, D.C.: WorldBank. Available online [April 26, 2004] at:www.worldbank.org/ogmc/files/237Kojim-831.pdf.


———. 2001b. “Emission Control: Privatizing Vehicle Inspectionand Reducing Fraud in Mexico City.” Viewpoint 238, Sep-tember. Washington, D.C.: World Bank. Available online[April 26, 2004] at: http://rru.worldbank.org/viewpoint/HTMLNotes/238/238summary.html.

Kojima, M., R.W. Bacon, J. Shah, M.S. Mainkar, M.K. Chaudhari,B. Bhanot, N.V. Iyer, A. Smith, and W.D. Atkinson. 2002.“Measurement of Mass Emissions from In-Use Two-StrokeEngine Three-Wheelers in South Asia.” SAE Technical Pa-per Series 2002-01-1681. Warrendale, Pennsylvania: SAE.

Kweon, Chol-Bum, D.E. Foster, J.S. Schauer, and S. Okada. 2002.“Detailed Chemical Composition and Particle Size Assess-ment of Diesel Engine Exhaust.” SAE Technical Paper2002-01-2670. Warrendale, Pennsylvania: SAE.

Larssen, Steinar, Frederick Gram, Leif Otto Hagen, Huib Jansen,Xander Olsthoorn, Reynaldo Lesaca, Emmanuel Anglo,Elma B. Torres, Ronald D. Subida, and Herminia A.Fransisco. 1996a. URBAIR Urban Air Quality ManagementStrategy in Asia: Metro Manila Report. Jitendra Shah andTanvi Nagpal, eds., October. Washington, D.C.: WorldBank. Available online [April 26, 2004] at: www-wds.worldbank.org/servlet/WDSServlet?pcont=details&eid=000009265_3980313101901.

Larssen, Steinar, Frederick Gram, Leif Otto Hagen, Huib Jansen,Xander Olsthoorn, Rajiv V. Aundhe, and Ulhas Joglekar.1996b. URBAIR Urban Air Quality Management Strategy inAsia: Greater Mumbai Report. Jitendra Shah and TanviNagpal, eds., October. Washington, D.C.: World Bank.Available online [April 26, 2004] at: www-wds.worldbank.org/servlet/WDSServlet?pcont=details&eid=000009265_3980217141523.

Lee, Rob, Joanna Pedley, and Christine Hobbs. 1998. “Fuel Qual-ity Impact on Heavy Duty Diesel Emissions: A LiteratureReview.” SAE Technical Paper Series 982649. Warrendale,Pennsylvania: SAE.

Lloyd, Alan C., and Thomas A. Cackette. 2001. “Diesel Engines:Environmental Impact and Control.” Journal of Air andWaste Management Association 51:809–847.

López, Gianni. 2002. “Air Quality Progress in Santiago.” Presenta-tion made at the first meeting of the Partners for CleanerFuels and Vehicles, United Nations Headquarters, NewYork City, November 14–15. Available online [April 26,2004] at: www.un.org/esa/gite/cleanfuels/chile.pdf.

Lowell, Dana. 2003. MTA (Metropolitan Transportation Authority)New York City Transit, Department of Buses. 2003. Per-sonal communication on their experience with operatingbuses equipped with continuously regenerating traps run-ning on 30 ppm sulfur diesel.

LPG Autofuel Systems. 2002. “Why Is LPG Not Used MoreWidely?” Available online [April 26, 2004] at:www.lpgautofuel.co.uk/what_is_lpg_.htm#uses.

Lvovsky, Kseniya. 2001. “Environment Strategy Papers No. 1:Health and Environment.” October. Washington, D.C.:World Bank. Available online [April 26, 2004] at: http://lnweb18.worldbank.org/ESSD/envext.nsf/41ByDocName/EnvironmentStrategyPaperNo1HealthandEnvironment2001361KBPDF/$FILE/ESP1Health&Environment2001.pdf.

Lvovsky, Kseniya, and Gordon Hughes. 1999. “EnvironmentalChallenges of Fuel Use.” Pollution Management In Focus Dis-cussion Note 7. Washington, D.C.: World Bank. Availableonline [April 26, 2004] at: http://wbln0018.worldbank.org/Networks/ESSD/icdb.nsf/D4856F112E805DF4852566C9007C27A6/2DD32C110D7A855085256885007A4287/$FILE/In+Focus+7.pdf.

McCarthy, Christopher I. 1994. “Update on the Effect of Govern-ment Regulations on Diesel Fuels.” Presentation made atthe 1994 NPRA (National Petrochemicals and Refiners As-sociation) National Fuels and Lubricants Meeting, Hous-ton, Texas, November 3–4.

MECA (Manufacturers of Emission Controls Association). 1998. “TheImpact of Gasoline Fuel Sulfur on Catalytic Emission ControlSystems.” September. Washington, D.C. Available online[April 26, 2004] at: http://meca.org/jahia/Jahia/pid/229.

———. 1999. “Demonstration of Advanced Emission ControlTechnologies Enabling Diesel-Powered Heavy-Duty En-gines to Achieve Low Emission Levels. Final Report.” Sep-tember. Washington, D.C. Available online [April 26, 2004]at: http://meca.org/jahia/Jahia/pid/229.

Mexico Air Quality Management Team. 2002. “Improving AirQuality in Metropolitan Mexico City: An Economic Valua-tion.” World Bank Research Working Paper 2785. Washing-ton, D.C.: World Bank. Available online [April 26, 2004] at:http://econ.worldbank.org/files/12030_wps2785.pdf.

Milkkonen, S., U. Kiiski, M. Mäkelä, A. Niemi, M. Niemi, L.Rantanen, and P. Saikkonen. 1997. “Reformulated DieselFuel - Four Years Experience in Finland.” SAE Paper971634. Warrendale, Pennsylvania: SAE.

Mok, W.C. 2001. “The Experience of Hong Kong SAR, China, inControlling Smoky Vehicles.” Available online [April 26,2004] at : www.worldbank.org/wbi/cleanair/caiasia/learningactivities/bangkok_documents/present11.ppt.

MTA (Metropolitan Transportation Authority) New York CityTransit Department of Buses. 2000. “NYTC Clean Fuel BusPrograms.” Presentation made at the Washington Metro-politan Area Transit Authority Alternative Fuels Workshop,July 6. Available online [April 26, 2004] at: www.epa.gov/otaq/retrofit/documents/nyc_2.pdf.

———. 2001. “Interim Report: Emissions Results from Clean Die-sel Demonstration Program with CRTTM Particulate Filterat New York City Transit.” Available online [April 26, 2004]at: www.epa.gov/OMS/retrofit/documents/nyc_crt_presentation.pdf.

NEPC (National Environment Protection Council). 2000. “Pro-posed Diesel Vehicle Emissions National Environment Pro-tection Measure Preparatory Work, In-Service EmissionsPerformance—Phase 2: Vehicle Testing.” Available online[April 26, 2004] at: www.ephc.gov.au/pdf/diesel/diesel_project2_2.pdf.

———. 2001a. “Proposed Diesel Vehicle Emissions National Envi-ronment Protection Measure Preparatory Work, In-ServiceCertification Correlation Studies.” Available online [April26, 2004] at: www.ephc.gov.au/pdf/diesel/Correlation_Studies.pdf.

———. 2001b. “Proposed Diesel Vehicle Emissions National Envi-ronment Protection Measure Preparatory Work, In-ServiceEmissions Testing—Pilot Study, Fault Identification and Ef-fect of Maintenance.” Available online [April 26, 2004] at:www.ephc.gov.au/pdf/diesel/PilotStudy.pdf.

Newbery, David, and Nicholas Stern, eds. 1987. The Theory of Taxationfor Developing Countries. New York: Oxford University Press.

Ntziachristos, Leonides, and Zissis Samaras. 2000. Technical Re-port No. 49, “COPERT III, Computer Programme to Calcu-late Emissions from Road, Transport Methodology andEmission Factors (Version 2.1),” November. Copenhagen:European Environment Agency. Available online [April 26,2004] at: http://reports.eea.eu.int/ Technical_report_No_49/en/tech49.pdf.

REFERENCES 167

OECD (Organisation for Economic Co-operation and Develop-ment). 1991. Mobility in Urban Areas: Learning to ConserveTransport. Paris.

Pandey, K.D., K. Bolt, U. Deichmann, K. Hamilton, B. Ostro, andD. Wheeler. 2003. “The Human Cost of Air Pollution: NewEstimates for Developing Countries.” World Bank, Devel-opment Research Group and Environment Department.Washington, D.C.: World Bank.

Park, S.S., M.S. Bae, and Y.J. Kim. 2001. “Chemical compositionand source apportionment of PM2.5 Particles in the SihwaArea, Korea.” Journal of Air and Waste Management Associa-tion 51(3): 393–405.

Pinto, J.P., R.K. Stevens, R.D. Willis, R.B. Kellog, Y. Mamane, J.Novak, J. Antroch, I. Bele, J. Leniek, and V. Vure. 1998.“Czech Air Quality Monitoring and Receptor ModelingStudy.” Environmental Science & Technology 32(7): 843–854.

Platts Oilgram Price Report. 2004. “EPA to Review Calif. Oxygen-ate Waiver Request.” February 2.

Pope, C.A., R.T. Burnett, M.J. Thun, E.E. Calle, D. Krewski, K. Ito,and G. Thurston. 2002. “Lung Cancer, CardiopulmonaryMortality, and Long-term Exposure to Fine Particulate AirPollution.” Journal of American Medical Association 287(9):1132–1141.

Radian. 1998. “PM Abatement Strategy for the Bangkok Metro-politan Area.” Table 4.3. Austin, Texas: Radian Interna-tional.

Rogers, John. 2002. “Assessment of the Pollution Under ControlProgram in India and Recommendations for Improve-ment.” A report prepared for the World Bank, October.Available online [April 26, 2004] at: www.worldbank.org/sarurbanair.

Sandmo, Agnar. 1975. “Optimal Taxation in the Presence of Exter-nalities.” Swedish Journal of Economics 77: 86–98.

Shu, Jet P.H. 2001. “Strategies to Reduce Two-Stroke Motorcycle inTaiwan.” Presentation made at the Regional Workshop forReducing Vehicle Emissions, Hanoi, Vietnam, September 6.Available online [April 26, 2004] at: www.adb.org/Docu-ments/Events/2001/RETA5937/Hanoi/documents/28_Shu.pdf.

Smith, Kirk. 1998. “National Burden of Disease from Indoor AirPollution in India.” Mumbai: Indira Gandhi Institute forDevelopment Research.

Thurston, George, and Kazuhiko Ito. 1999. “EpidemiologicalStudies of Ozone Exposure Effects.” In S.T. Holgate, J. M.Samet, H.S. Kore, and R. L. Maynard, eds., Air Pollution andHealth. London: Academic Press.

Tsang, Matthew, and Kong Ha. 2002. “Hong Kong Experience inRetrofitting In-Use Diesel Vehicles.” Paper presented atBetter Air Quality 2002, Regional Workshop on Better AirQuality in Asia and Pacific Rim Cities, Hong Kong, China,December 16–18. Available online [April 26, 2004] at:www.cse.polyu.edu.hk/~activi/BAQ2002/BAQ2002_files/Proceedings/Subworkshop4/sw4a-4Tsang_paper.pdf.

UK Commission for Integrated Transport. “Local Air Quality, Cli-mate Change.” Available online [April 26, 2004] at:www.cfit.gov.uk/mf/reports/laq/.

UNECE (United Nations Economic Commission for Europe).2002. “Transport Division. World Forum for Harmoniza-tion of Vehicle Regulations.” Available online [April 26,2004] at: www.unece.org/trans/main/wp29/wp29regs.html.

UOP. 1994. “The Challenge of Reformulated Gasoline: An Updateon the Clean Air Act and the Refining Industry.” DesPlaines, Illinois.

USDA (U.S. Department of Agriculture). 2002. “Agricultural Out-look/January-February 2002, Statistical Indicators.” Avail-able online [April 26, 2004] at: www.ers.usda.gov/publica-tions/agoutlook/jan2002/ao288h.pdf.

———. 2003. “Brazilian Sugar.” October. Available online [April26, 2004] at: www.fas.usda.gov/htp/sugar/2003/Brazilsugar03.pdf.

U.S. EPA (Environmental Protection Agency). 1997a. “EmissionStandards Reference Guide for Heavy-Duty and Nonroad En-gines.” EPA420-F97-014, September. Available online [April 26,2004] at: www.epa.gov/otaq/cert/hd-cert/stds-eng.pdf.

———. 1997b. “The Benefits and Costs of the Clean Air Act, 1970to 1990.” Available online [April 26, 2004] at:www.epa.gov/oar/sect812/copy.html.

———. 1999a. “Regulatory Impact Analysis, Chapter VII: Benefit-Cost Analysis.” EPA 420-R-99-023, December 22. Availableonline [April 26, 2004] at: www.epa.gov/otaq/regs/ld-hwy/tier-2/frm/ria/chvii.pdf.

———. 1999b. “Heavy-Duty Diesel Emission Reduction ProjectRetrofit/Rebuild Component.” EPA420-R-99-014, June.Available online [April 26, 2004] at: www.epa.gov/otaq/retrofit/documents/epafinalrep.pdf.

———. 1999c. “Phase II Reformulated Gasoline: the next majorstep toward cleaner air.” EPA420-F-99-042, November.Available online [April 26, 2004] at: www.epa.gov/otaq/eparfg.pdf.

———. 1999d. “Regulatory Announcement: EPA’s Program forCleaner Vehicles and Cleaner Gasoline.” EPQ420-F-99-051,December. Available online [April 26, 2004] at:www.epa.gov/otaq/regs/ld-hwy/tier-2/frm/f99051.pdf.

———. 1999e. “Regulatory Impact Analysis, Chapter V: EconomicImpact.” EPA 420-R-99-023, December 22. Available online[April 26, 2004] at: www.epa.gov/otaq/regs/ld-hwy/tier-2/frm/ria/chv.pdf.

———. 1999f. “The Benefits and Costs of the Clean Air Act, 1990to 2010.” EPA-410-R-99-001, Office of Air Radiation andOffice of Policy, November. Available online [April 26,2004] at: www.epa.gov/oar/sect812/.

———. 2000a. “Final Emission Standards for 2004 and LaterModel Year Highway Heavy-Duty Vehicles and Engines.”EPA420-F-00-026, July. Available online [April 26, 2004] at:www.epa.gov/OMS/regs/hd-hwy/2000frm/f00026.htm.

———. 2000b. “Federal and California Exhaust and EvaporativeEmission Standards for Light-Duty Vehicles and Light-Duty Trucks.” EPA420-B-00-001, February. Available online[April 26, 2004] at: www.epa.gov/otaq/cert/veh-cert/b00001.xls.

———. 2000c. Federal Register Vol. 65, No. 28, February 10. Avail-able online [April 26, 2004] at: www.epa.gov/otaq/regs/ld-hwy/tier-2/frm/fr-t2reg.pdf.

———. 2000d. “Heavy-Duty Engine and Vehicle Standards andHighway Diesel Fuel Sulfur Control Requirements.”EPA420-F-00-057. Available online [April 26, 2004] at:www.epa.gov/OMS/regs/hd2007/frm/f00057.htm.

———. 2003. “EPA Finalizes Emission Standards for New High-way Motorcycles.” EPA420-F-03-044, December. Availableonline [April 26, 2004] at: www.epa.gov/otaq/regs/roadbike/420f03044.pdf.


———. Undated a. “Mobile Source Emissions—Past, Present, andFuture.” Available online [April 26, 2004] at:www.epa.gov/otaq/invntory/overview/solutions/milestones.htm.

———. Undated b. “Voluntary Diesel Retrofit Program. EmissionsStandards Timeline.” Available online [April 26, 2004] at:www.epa.gov/otaq/retrofit/overoh-all.htm.

U.S. GAO (General Accounting Office). 2000. “Petroleum andEthanol Fuels: Tax Incentives and Related GAO Work.”GAO/RCED-00-301R Tax Incentives for Petroleum andEthanol Fuels, Washington, D.C. Available online [April 26,2004] at: www.gao.gov/archive/2000/rc00301r.pdf.

———. 2004. “Air Pollution: EPA Could Take Additional Steps toHelp Maximize the Benefits from the 2007 Diesel EmissionsStandards.” GAO–04–313, Washington, D.C. Available online[April 26, 2004] at www.gao.gov/new.items/d04313.pdf.

Vega, E., I. Garcîa, D. Apam, M.E. Ruîz, and M. Barbiaux. 1997.“Application of a Chemical Mass Balance Receptor Modelto Respirable Particulate Matter in Mexico City.” Journal ofAir and Waste Management Association 47(4): 524–529.

Watson, John G., Eric Fujita, Judith C. Chow, Barbara Zielinska, L.Willard Richards, William Neff, and David Dietrich. 1998.“Northern Front Range Air Quality Study Final Report.”DRI Document No. 6580-685-8750.1F2. Available online[April 26, 2004] at: www.nfraqs.colostate.edu/.

Watt, Glen M. 2001. “Natural Gas Vehicle Transit Bus Fleets: TheCurrent International Experience.” IANGV (InternationalAssociation for Natural Gas Vehicles) Review Paper. Avail-able online [April 26, 2004] at: www.iangv.org/html/sources/sources/reports/iangv_bus_report.pdf.

Weaver, Christopher. 2003. Engine, Fuel, and Emissions Engineer-ing, Inc. Personal communication.

Well, Peter. 2003. “Jam Tomorrow? The Environmental Cost ofWaiting for Fuel Cell Cars.” Automotive Environment Ana-lyst 95: 23–25.

WHO (World Health Organization). 2000. Guidelines for Air Qual-ity. Geneva: WHO. Available online [April 26, 2004] at:www.euro.who.int/air/Activities/20020620_1.

———. 2002. “Attributable Mortality by Risk Factor, Level of De-velopment and Sex, 2000.” Annex Table 9 in The WorldHealth Report 2002. Geneva. Available online [April 26,2004] at: www.who.int/whr/2002/en/whr2002_annex9_10.pdf.

Working Party on National Environmental Policy, Working Group onTransport. 2002. “Policy Instruments for Achieving Environ-mentally Sustainable Transport: Report on Phase III of theProject on Environmentally Sustainable Transport.” OECD,April. ENV/EPOC/WPNEP/T(2001)7/FINAL. Availableonline [April 26, 2004] at: www.olis.oecd.org/olis/2001doc.nsf/LinkTo/ENV-EPOC-WPNEP-T(2001)7-FINAL.

World Bank. 1994. “Chile: Managing Environmental Problems:Economic Analysis of Selected Issues.” Report 13061-CH.Washington, D.C.

———. 1999. Pollution Prevention and Abatement Handbook 1998:Toward Cleaner Production. Washington, D.C. Availableonline [April 26, 2004] at: http://lnweb18.worldbank.org/ESSD/envext.nsf/51ByDocName/PollutionPreventionandAbatementHandbook.

———. 2001. “Urban Air Pollution: How Can Urban Bus PolicyReduce Air Pollution?” South Asia Urban Air Quality Man-agement Briefing Note No. 3. Washington, D.C. Availableonline [April 26, 2004] at: www.worldbank.org/sarurbanair.

———. 2002a. “Urban Air Pollution: What Do We Know AboutAir Pollution?—India Case Study.” South Asia Urban AirQuality Management Briefing Note No. 4. Washington,D.C. Available online [April 26, 2004] at:www.worldbank.org/sarurbanair.

———. 2002b. “Urban Air Pollution: Impact of Better Traffic Man-agement.” South Asia Urban Air Quality ManagementBriefing Note No. 5. Washington, D.C. Available online[April 26, 2004] at: www.worldbank.org/sarurbanair.

———. 2002c. “Urban Air Pollution: Urban Planning and AirQuality.” South Asia Urban Air Quality Management Brief-ing Note No. 6. Washington, D.C. Available online [April26, 2004] at: www.worldbank.org/sarurbanair.

———. 2002d. “Urban Air Pollution: Catching Gasoline and Die-sel Adulteration.” South Asia Urban Air Quality Manage-ment Briefing Note No. 7. Washington, D.C. Availableonline [April 26, 2004] at: www.worldbank.org/sarurbanair.

———. 2002e. “Urban Air Pollution: Can Vehicle Scrappage Pro-grams Be Successful?” South Asia Urban Air Quality Man-agement Briefing Note No. 8. Washington, D.C. Availableonline [April 26, 2004] at: www.worldbank.org/sarurbanair.

———. 2002f. Chapter 11, “Strengthening Urban Transport Insti-tutions.” In Cities on the Move: A World Bank Urban TransportStrategy Review. Washington, D.C. Available online [April26, 2004] at: www.worldbank.org/transport/utsr/all_chap.pdf.

———. 2003a. “Urban Air Pollution: Health Impacts of OutdoorAir Pollution.” South Asia Urban Air Quality ManagementBriefing Note No. 11, February. Washington, D.C. Availableonline [April 26, 2004] at: www.worldbank.org/sarurbanair.

———. 2003b. “Urban Air Pollution: Economic Valuation of theHealth Benefits of Reduction in Air Pollution.” South AsiaUrban Air Quality Management Briefing Note No. 12, Feb-ruary. Washington, D.C. Available online [April 26, 2004]at: www.worldbank.org/sarurbanair.

———. 2003c. “Urban Air Pollution: The Science of Health Im-pacts of Particulate Matter.” South Asia Urban Air QualityManagement Briefing Note No. 13, March. Washington,D.C. Available online [April 26, 2004] at:www.worldbank.org/sarurbanair.

Yamaguchi, Nancy. 2003. Trans-Energy. 2003. Personal communi-cation.

Yamaguchi, Nancy, David Fridley, and Ke Xiaoming. 2002. “Im-proving Transport Fuel Quality in China: Implications forthe Refining Sector.” Available online [April 26, 2004] at:http://eetd.lbl.gov/ea/China/pubs/china_refining_study_final.pdf.